1 @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,
2 @c 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
8 @chapter Target Description Macros and Functions
9 @cindex machine description macros
10 @cindex target description macros
11 @cindex macros, target description
12 @cindex @file{tm.h} macros
14 In addition to the file @file{@var{machine}.md}, a machine description
15 includes a C header file conventionally given the name
16 @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}.
17 The header file defines numerous macros that convey the information
18 about the target machine that does not fit into the scheme of the
19 @file{.md} file. The file @file{tm.h} should be a link to
20 @file{@var{machine}.h}. The header file @file{config.h} includes
21 @file{tm.h} and most compiler source files include @file{config.h}. The
22 source file defines a variable @code{targetm}, which is a structure
23 containing pointers to functions and data relating to the target
24 machine. @file{@var{machine}.c} should also contain their definitions,
25 if they are not defined elsewhere in GCC, and other functions called
26 through the macros defined in the @file{.h} file.
29 * Target Structure:: The @code{targetm} variable.
30 * Driver:: Controlling how the driver runs the compilation passes.
31 * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}.
32 * Per-Function Data:: Defining data structures for per-function information.
33 * Storage Layout:: Defining sizes and alignments of data.
34 * Type Layout:: Defining sizes and properties of basic user data types.
35 * Registers:: Naming and describing the hardware registers.
36 * Register Classes:: Defining the classes of hardware registers.
37 * Old Constraints:: The old way to define machine-specific constraints.
38 * Stack and Calling:: Defining which way the stack grows and by how much.
39 * Varargs:: Defining the varargs macros.
40 * Trampolines:: Code set up at run time to enter a nested function.
41 * Library Calls:: Controlling how library routines are implicitly called.
42 * Addressing Modes:: Defining addressing modes valid for memory operands.
43 * Anchored Addresses:: Defining how @option{-fsection-anchors} should work.
44 * Condition Code:: Defining how insns update the condition code.
45 * Costs:: Defining relative costs of different operations.
46 * Scheduling:: Adjusting the behavior of the instruction scheduler.
47 * Sections:: Dividing storage into text, data, and other sections.
48 * PIC:: Macros for position independent code.
49 * Assembler Format:: Defining how to write insns and pseudo-ops to output.
50 * Debugging Info:: Defining the format of debugging output.
51 * Floating Point:: Handling floating point for cross-compilers.
52 * Mode Switching:: Insertion of mode-switching instructions.
53 * Target Attributes:: Defining target-specific uses of @code{__attribute__}.
54 * Emulated TLS:: Emulated TLS support.
55 * MIPS Coprocessors:: MIPS coprocessor support and how to customize it.
56 * PCH Target:: Validity checking for precompiled headers.
57 * C++ ABI:: Controlling C++ ABI changes.
58 * Named Address Spaces:: Adding support for named address spaces
59 * Misc:: Everything else.
62 @node Target Structure
63 @section The Global @code{targetm} Variable
65 @cindex target functions
67 @deftypevar {struct gcc_target} targetm
68 The target @file{.c} file must define the global @code{targetm} variable
69 which contains pointers to functions and data relating to the target
70 machine. The variable is declared in @file{target.h};
71 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
72 used to initialize the variable, and macros for the default initializers
73 for elements of the structure. The @file{.c} file should override those
74 macros for which the default definition is inappropriate. For example:
77 #include "target-def.h"
79 /* @r{Initialize the GCC target structure.} */
81 #undef TARGET_COMP_TYPE_ATTRIBUTES
82 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
84 struct gcc_target targetm = TARGET_INITIALIZER;
88 Where a macro should be defined in the @file{.c} file in this manner to
89 form part of the @code{targetm} structure, it is documented below as a
90 ``Target Hook'' with a prototype. Many macros will change in future
91 from being defined in the @file{.h} file to being part of the
92 @code{targetm} structure.
95 @section Controlling the Compilation Driver, @file{gcc}
97 @cindex controlling the compilation driver
99 @c prevent bad page break with this line
100 You can control the compilation driver.
102 @defmac SWITCH_TAKES_ARG (@var{char})
103 A C expression which determines whether the option @option{-@var{char}}
104 takes arguments. The value should be the number of arguments that
105 option takes--zero, for many options.
107 By default, this macro is defined as
108 @code{DEFAULT_SWITCH_TAKES_ARG}, which handles the standard options
109 properly. You need not define @code{SWITCH_TAKES_ARG} unless you
110 wish to add additional options which take arguments. Any redefinition
111 should call @code{DEFAULT_SWITCH_TAKES_ARG} and then check for
115 @defmac WORD_SWITCH_TAKES_ARG (@var{name})
116 A C expression which determines whether the option @option{-@var{name}}
117 takes arguments. The value should be the number of arguments that
118 option takes--zero, for many options. This macro rather than
119 @code{SWITCH_TAKES_ARG} is used for multi-character option names.
121 By default, this macro is defined as
122 @code{DEFAULT_WORD_SWITCH_TAKES_ARG}, which handles the standard options
123 properly. You need not define @code{WORD_SWITCH_TAKES_ARG} unless you
124 wish to add additional options which take arguments. Any redefinition
125 should call @code{DEFAULT_WORD_SWITCH_TAKES_ARG} and then check for
129 @defmac SWITCH_CURTAILS_COMPILATION (@var{char})
130 A C expression which determines whether the option @option{-@var{char}}
131 stops compilation before the generation of an executable. The value is
132 boolean, nonzero if the option does stop an executable from being
133 generated, zero otherwise.
135 By default, this macro is defined as
136 @code{DEFAULT_SWITCH_CURTAILS_COMPILATION}, which handles the standard
137 options properly. You need not define
138 @code{SWITCH_CURTAILS_COMPILATION} unless you wish to add additional
139 options which affect the generation of an executable. Any redefinition
140 should call @code{DEFAULT_SWITCH_CURTAILS_COMPILATION} and then check
141 for additional options.
144 @defmac TARGET_OPTION_TRANSLATE_TABLE
145 If defined, a list of pairs of strings, the first of which is a
146 potential command line target to the @file{gcc} driver program, and the
147 second of which is a space-separated (tabs and other whitespace are not
148 supported) list of options with which to replace the first option. The
149 target defining this list is responsible for assuring that the results
150 are valid. Replacement options may not be the @code{--opt} style, they
151 must be the @code{-opt} style. It is the intention of this macro to
152 provide a mechanism for substitution that affects the multilibs chosen,
153 such as one option that enables many options, some of which select
154 multilibs. Example nonsensical definition, where @option{-malt-abi},
155 @option{-EB}, and @option{-mspoo} cause different multilibs to be chosen:
158 #define TARGET_OPTION_TRANSLATE_TABLE \
159 @{ "-fast", "-march=fast-foo -malt-abi -I/usr/fast-foo" @}, \
160 @{ "-compat", "-EB -malign=4 -mspoo" @}
164 @defmac DRIVER_SELF_SPECS
165 A list of specs for the driver itself. It should be a suitable
166 initializer for an array of strings, with no surrounding braces.
168 The driver applies these specs to its own command line between loading
169 default @file{specs} files (but not command-line specified ones) and
170 choosing the multilib directory or running any subcommands. It
171 applies them in the order given, so each spec can depend on the
172 options added by earlier ones. It is also possible to remove options
173 using @samp{%<@var{option}} in the usual way.
175 This macro can be useful when a port has several interdependent target
176 options. It provides a way of standardizing the command line so
177 that the other specs are easier to write.
179 Do not define this macro if it does not need to do anything.
182 @defmac OPTION_DEFAULT_SPECS
183 A list of specs used to support configure-time default options (i.e.@:
184 @option{--with} options) in the driver. It should be a suitable initializer
185 for an array of structures, each containing two strings, without the
186 outermost pair of surrounding braces.
188 The first item in the pair is the name of the default. This must match
189 the code in @file{config.gcc} for the target. The second item is a spec
190 to apply if a default with this name was specified. The string
191 @samp{%(VALUE)} in the spec will be replaced by the value of the default
192 everywhere it occurs.
194 The driver will apply these specs to its own command line between loading
195 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
196 the same mechanism as @code{DRIVER_SELF_SPECS}.
198 Do not define this macro if it does not need to do anything.
202 A C string constant that tells the GCC driver program options to
203 pass to CPP@. It can also specify how to translate options you
204 give to GCC into options for GCC to pass to the CPP@.
206 Do not define this macro if it does not need to do anything.
209 @defmac CPLUSPLUS_CPP_SPEC
210 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
211 than C@. If you do not define this macro, then the value of
212 @code{CPP_SPEC} (if any) will be used instead.
216 A C string constant that tells the GCC driver program options to
217 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
219 It can also specify how to translate options you give to GCC into options
220 for GCC to pass to front ends.
222 Do not define this macro if it does not need to do anything.
226 A C string constant that tells the GCC driver program options to
227 pass to @code{cc1plus}. It can also specify how to translate options you
228 give to GCC into options for GCC to pass to the @code{cc1plus}.
230 Do not define this macro if it does not need to do anything.
231 Note that everything defined in CC1_SPEC is already passed to
232 @code{cc1plus} so there is no need to duplicate the contents of
233 CC1_SPEC in CC1PLUS_SPEC@.
237 A C string constant that tells the GCC driver program options to
238 pass to the assembler. It can also specify how to translate options
239 you give to GCC into options for GCC to pass to the assembler.
240 See the file @file{sun3.h} for an example of this.
242 Do not define this macro if it does not need to do anything.
245 @defmac ASM_FINAL_SPEC
246 A C string constant that tells the GCC driver program how to
247 run any programs which cleanup after the normal assembler.
248 Normally, this is not needed. See the file @file{mips.h} for
251 Do not define this macro if it does not need to do anything.
254 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
255 Define this macro, with no value, if the driver should give the assembler
256 an argument consisting of a single dash, @option{-}, to instruct it to
257 read from its standard input (which will be a pipe connected to the
258 output of the compiler proper). This argument is given after any
259 @option{-o} option specifying the name of the output file.
261 If you do not define this macro, the assembler is assumed to read its
262 standard input if given no non-option arguments. If your assembler
263 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
264 see @file{mips.h} for instance.
268 A C string constant that tells the GCC driver program options to
269 pass to the linker. It can also specify how to translate options you
270 give to GCC into options for GCC to pass to the linker.
272 Do not define this macro if it does not need to do anything.
276 Another C string constant used much like @code{LINK_SPEC}. The difference
277 between the two is that @code{LIB_SPEC} is used at the end of the
278 command given to the linker.
280 If this macro is not defined, a default is provided that
281 loads the standard C library from the usual place. See @file{gcc.c}.
285 Another C string constant that tells the GCC driver program
286 how and when to place a reference to @file{libgcc.a} into the
287 linker command line. This constant is placed both before and after
288 the value of @code{LIB_SPEC}.
290 If this macro is not defined, the GCC driver provides a default that
291 passes the string @option{-lgcc} to the linker.
294 @defmac REAL_LIBGCC_SPEC
295 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
296 @code{LIBGCC_SPEC} is not directly used by the driver program but is
297 instead modified to refer to different versions of @file{libgcc.a}
298 depending on the values of the command line flags @option{-static},
299 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
300 targets where these modifications are inappropriate, define
301 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
302 driver how to place a reference to @file{libgcc} on the link command
303 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
306 @defmac USE_LD_AS_NEEDED
307 A macro that controls the modifications to @code{LIBGCC_SPEC}
308 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
309 generated that uses --as-needed and the shared libgcc in place of the
310 static exception handler library, when linking without any of
311 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
315 If defined, this C string constant is added to @code{LINK_SPEC}.
316 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
317 the modifications to @code{LIBGCC_SPEC} mentioned in
318 @code{REAL_LIBGCC_SPEC}.
321 @defmac STARTFILE_SPEC
322 Another C string constant used much like @code{LINK_SPEC}. The
323 difference between the two is that @code{STARTFILE_SPEC} is used at
324 the very beginning of the command given to the linker.
326 If this macro is not defined, a default is provided that loads the
327 standard C startup file from the usual place. See @file{gcc.c}.
331 Another C string constant used much like @code{LINK_SPEC}. The
332 difference between the two is that @code{ENDFILE_SPEC} is used at
333 the very end of the command given to the linker.
335 Do not define this macro if it does not need to do anything.
338 @defmac THREAD_MODEL_SPEC
339 GCC @code{-v} will print the thread model GCC was configured to use.
340 However, this doesn't work on platforms that are multilibbed on thread
341 models, such as AIX 4.3. On such platforms, define
342 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
343 blanks that names one of the recognized thread models. @code{%*}, the
344 default value of this macro, will expand to the value of
345 @code{thread_file} set in @file{config.gcc}.
348 @defmac SYSROOT_SUFFIX_SPEC
349 Define this macro to add a suffix to the target sysroot when GCC is
350 configured with a sysroot. This will cause GCC to search for usr/lib,
351 et al, within sysroot+suffix.
354 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
355 Define this macro to add a headers_suffix to the target sysroot when
356 GCC is configured with a sysroot. This will cause GCC to pass the
357 updated sysroot+headers_suffix to CPP, causing it to search for
358 usr/include, et al, within sysroot+headers_suffix.
362 Define this macro to provide additional specifications to put in the
363 @file{specs} file that can be used in various specifications like
366 The definition should be an initializer for an array of structures,
367 containing a string constant, that defines the specification name, and a
368 string constant that provides the specification.
370 Do not define this macro if it does not need to do anything.
372 @code{EXTRA_SPECS} is useful when an architecture contains several
373 related targets, which have various @code{@dots{}_SPECS} which are similar
374 to each other, and the maintainer would like one central place to keep
377 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
378 define either @code{_CALL_SYSV} when the System V calling sequence is
379 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
382 The @file{config/rs6000/rs6000.h} target file defines:
385 #define EXTRA_SPECS \
386 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
388 #define CPP_SYS_DEFAULT ""
391 The @file{config/rs6000/sysv.h} target file defines:
395 "%@{posix: -D_POSIX_SOURCE @} \
396 %@{mcall-sysv: -D_CALL_SYSV @} \
397 %@{!mcall-sysv: %(cpp_sysv_default) @} \
398 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
400 #undef CPP_SYSV_DEFAULT
401 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
404 while the @file{config/rs6000/eabiaix.h} target file defines
405 @code{CPP_SYSV_DEFAULT} as:
408 #undef CPP_SYSV_DEFAULT
409 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
413 @defmac LINK_LIBGCC_SPECIAL_1
414 Define this macro if the driver program should find the library
415 @file{libgcc.a}. If you do not define this macro, the driver program will pass
416 the argument @option{-lgcc} to tell the linker to do the search.
419 @defmac LINK_GCC_C_SEQUENCE_SPEC
420 The sequence in which libgcc and libc are specified to the linker.
421 By default this is @code{%G %L %G}.
424 @defmac LINK_COMMAND_SPEC
425 A C string constant giving the complete command line need to execute the
426 linker. When you do this, you will need to update your port each time a
427 change is made to the link command line within @file{gcc.c}. Therefore,
428 define this macro only if you need to completely redefine the command
429 line for invoking the linker and there is no other way to accomplish
430 the effect you need. Overriding this macro may be avoidable by overriding
431 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
434 @defmac LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
435 A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search
436 directories from linking commands. Do not give it a nonzero value if
437 removing duplicate search directories changes the linker's semantics.
440 @defmac MULTILIB_DEFAULTS
441 Define this macro as a C expression for the initializer of an array of
442 string to tell the driver program which options are defaults for this
443 target and thus do not need to be handled specially when using
444 @code{MULTILIB_OPTIONS}.
446 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
447 the target makefile fragment or if none of the options listed in
448 @code{MULTILIB_OPTIONS} are set by default.
449 @xref{Target Fragment}.
452 @defmac RELATIVE_PREFIX_NOT_LINKDIR
453 Define this macro to tell @command{gcc} that it should only translate
454 a @option{-B} prefix into a @option{-L} linker option if the prefix
455 indicates an absolute file name.
458 @defmac MD_EXEC_PREFIX
459 If defined, this macro is an additional prefix to try after
460 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
461 when the compiler is built as a cross
462 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
463 to the list of directories used to find the assembler in @file{configure.in}.
466 @defmac STANDARD_STARTFILE_PREFIX
467 Define this macro as a C string constant if you wish to override the
468 standard choice of @code{libdir} as the default prefix to
469 try when searching for startup files such as @file{crt0.o}.
470 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
471 is built as a cross compiler.
474 @defmac STANDARD_STARTFILE_PREFIX_1
475 Define this macro as a C string constant if you wish to override the
476 standard choice of @code{/lib} as a prefix to try after the default prefix
477 when searching for startup files such as @file{crt0.o}.
478 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
479 is built as a cross compiler.
482 @defmac STANDARD_STARTFILE_PREFIX_2
483 Define this macro as a C string constant if you wish to override the
484 standard choice of @code{/lib} as yet another prefix to try after the
485 default prefix when searching for startup files such as @file{crt0.o}.
486 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
487 is built as a cross compiler.
490 @defmac MD_STARTFILE_PREFIX
491 If defined, this macro supplies an additional prefix to try after the
492 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
493 compiler is built as a cross compiler.
496 @defmac MD_STARTFILE_PREFIX_1
497 If defined, this macro supplies yet another prefix to try after the
498 standard prefixes. It is not searched when the compiler is built as a
502 @defmac INIT_ENVIRONMENT
503 Define this macro as a C string constant if you wish to set environment
504 variables for programs called by the driver, such as the assembler and
505 loader. The driver passes the value of this macro to @code{putenv} to
506 initialize the necessary environment variables.
509 @defmac LOCAL_INCLUDE_DIR
510 Define this macro as a C string constant if you wish to override the
511 standard choice of @file{/usr/local/include} as the default prefix to
512 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
513 comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
515 Cross compilers do not search either @file{/usr/local/include} or its
519 @defmac SYSTEM_INCLUDE_DIR
520 Define this macro as a C string constant if you wish to specify a
521 system-specific directory to search for header files before the standard
522 directory. @code{SYSTEM_INCLUDE_DIR} comes before
523 @code{STANDARD_INCLUDE_DIR} in the search order.
525 Cross compilers do not use this macro and do not search the directory
529 @defmac STANDARD_INCLUDE_DIR
530 Define this macro as a C string constant if you wish to override the
531 standard choice of @file{/usr/include} as the default prefix to
532 try when searching for header files.
534 Cross compilers ignore this macro and do not search either
535 @file{/usr/include} or its replacement.
538 @defmac STANDARD_INCLUDE_COMPONENT
539 The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}.
540 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
541 If you do not define this macro, no component is used.
544 @defmac INCLUDE_DEFAULTS
545 Define this macro if you wish to override the entire default search path
546 for include files. For a native compiler, the default search path
547 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
548 @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and
549 @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
550 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
551 and specify private search areas for GCC@. The directory
552 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
554 The definition should be an initializer for an array of structures.
555 Each array element should have four elements: the directory name (a
556 string constant), the component name (also a string constant), a flag
557 for C++-only directories,
558 and a flag showing that the includes in the directory don't need to be
559 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
560 the array with a null element.
562 The component name denotes what GNU package the include file is part of,
563 if any, in all uppercase letters. For example, it might be @samp{GCC}
564 or @samp{BINUTILS}. If the package is part of a vendor-supplied
565 operating system, code the component name as @samp{0}.
567 For example, here is the definition used for VAX/VMS:
570 #define INCLUDE_DEFAULTS \
572 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
573 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
574 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
581 Here is the order of prefixes tried for exec files:
585 Any prefixes specified by the user with @option{-B}.
588 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
589 is not set and the compiler has not been installed in the configure-time
590 @var{prefix}, the location in which the compiler has actually been installed.
593 The directories specified by the environment variable @code{COMPILER_PATH}.
596 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
597 in the configured-time @var{prefix}.
600 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
603 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
606 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
610 Here is the order of prefixes tried for startfiles:
614 Any prefixes specified by the user with @option{-B}.
617 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
618 value based on the installed toolchain location.
621 The directories specified by the environment variable @code{LIBRARY_PATH}
622 (or port-specific name; native only, cross compilers do not use this).
625 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
626 in the configured @var{prefix} or this is a native compiler.
629 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
632 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
636 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
637 native compiler, or we have a target system root.
640 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
641 native compiler, or we have a target system root.
644 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
645 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
646 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
649 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
650 compiler, or we have a target system root. The default for this macro is
654 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
655 compiler, or we have a target system root. The default for this macro is
659 @node Run-time Target
660 @section Run-time Target Specification
661 @cindex run-time target specification
662 @cindex predefined macros
663 @cindex target specifications
665 @c prevent bad page break with this line
666 Here are run-time target specifications.
668 @defmac TARGET_CPU_CPP_BUILTINS ()
669 This function-like macro expands to a block of code that defines
670 built-in preprocessor macros and assertions for the target CPU, using
671 the functions @code{builtin_define}, @code{builtin_define_std} and
672 @code{builtin_assert}. When the front end
673 calls this macro it provides a trailing semicolon, and since it has
674 finished command line option processing your code can use those
677 @code{builtin_assert} takes a string in the form you pass to the
678 command-line option @option{-A}, such as @code{cpu=mips}, and creates
679 the assertion. @code{builtin_define} takes a string in the form
680 accepted by option @option{-D} and unconditionally defines the macro.
682 @code{builtin_define_std} takes a string representing the name of an
683 object-like macro. If it doesn't lie in the user's namespace,
684 @code{builtin_define_std} defines it unconditionally. Otherwise, it
685 defines a version with two leading underscores, and another version
686 with two leading and trailing underscores, and defines the original
687 only if an ISO standard was not requested on the command line. For
688 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
689 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
690 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
691 defines only @code{_ABI64}.
693 You can also test for the C dialect being compiled. The variable
694 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
695 or @code{clk_objective_c}. Note that if we are preprocessing
696 assembler, this variable will be @code{clk_c} but the function-like
697 macro @code{preprocessing_asm_p()} will return true, so you might want
698 to check for that first. If you need to check for strict ANSI, the
699 variable @code{flag_iso} can be used. The function-like macro
700 @code{preprocessing_trad_p()} can be used to check for traditional
704 @defmac TARGET_OS_CPP_BUILTINS ()
705 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
706 and is used for the target operating system instead.
709 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
710 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
711 and is used for the target object format. @file{elfos.h} uses this
712 macro to define @code{__ELF__}, so you probably do not need to define
716 @deftypevar {extern int} target_flags
717 This variable is declared in @file{options.h}, which is included before
718 any target-specific headers.
721 @hook TARGET_DEFAULT_TARGET_FLAGS
722 This variable specifies the initial value of @code{target_flags}.
723 Its default setting is 0.
726 @cindex optional hardware or system features
727 @cindex features, optional, in system conventions
729 @hook TARGET_HANDLE_OPTION
730 This hook is called whenever the user specifies one of the
731 target-specific options described by the @file{.opt} definition files
732 (@pxref{Options}). It has the opportunity to do some option-specific
733 processing and should return true if the option is valid. The default
734 definition does nothing but return true.
736 @var{code} specifies the @code{OPT_@var{name}} enumeration value
737 associated with the selected option; @var{name} is just a rendering of
738 the option name in which non-alphanumeric characters are replaced by
739 underscores. @var{arg} specifies the string argument and is null if
740 no argument was given. If the option is flagged as a @code{UInteger}
741 (@pxref{Option properties}), @var{value} is the numeric value of the
742 argument. Otherwise @var{value} is 1 if the positive form of the
743 option was used and 0 if the ``no-'' form was.
746 @hook TARGET_HANDLE_C_OPTION
747 This target hook is called whenever the user specifies one of the
748 target-specific C language family options described by the @file{.opt}
749 definition files(@pxref{Options}). It has the opportunity to do some
750 option-specific processing and should return true if the option is
751 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
752 default definition does nothing but return false.
754 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
755 options. However, if processing an option requires routines that are
756 only available in the C (and related language) front ends, then you
757 should use @code{TARGET_HANDLE_C_OPTION} instead.
760 @defmac TARGET_VERSION
761 This macro is a C statement to print on @code{stderr} a string
762 describing the particular machine description choice. Every machine
763 description should define @code{TARGET_VERSION}. For example:
767 #define TARGET_VERSION \
768 fprintf (stderr, " (68k, Motorola syntax)");
770 #define TARGET_VERSION \
771 fprintf (stderr, " (68k, MIT syntax)");
776 @defmac OVERRIDE_OPTIONS
777 Sometimes certain combinations of command options do not make sense on
778 a particular target machine. You can define a macro
779 @code{OVERRIDE_OPTIONS} to take account of this. This macro, if
780 defined, is executed once just after all the command options have been
783 Don't use this macro to turn on various extra optimizations for
784 @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
786 If you need to do something whenever the optimization level is
787 changed via the optimize attribute or pragma, see
788 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}
790 This macros is obsolete, new ports should use the target hook
791 @code{TARGET_OPTION_OVERRIDE} instead.
794 @hook TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE
795 This target function is similar to the hook @code{TARGET_OPTION_OVERRIDE}
796 but is called when the optimize level is changed via an attribute or
797 pragma or when it is reset at the end of the code affected by the
798 attribute or pragma. It is not called at the beginning of compilation
799 when @code{TARGET_OPTION_OVERRIDE} is called so if you want to perform these
800 actions then, you should have @code{TARGET_OPTION_OVERRIDE} call
801 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}.
804 @defmac C_COMMON_OVERRIDE_OPTIONS
805 This is similar to @code{OVERRIDE_OPTIONS} but is only used in the C
806 language frontends (C, Objective-C, C++, Objective-C++) and so can be
807 used to alter option flag variables which only exist in those
811 @defmac OPTIMIZATION_OPTIONS (@var{level}, @var{size})
812 Some machines may desire to change what optimizations are performed for
813 various optimization levels. This macro, if defined, is executed once
814 just after the optimization level is determined and before the remainder
815 of the command options have been parsed. Values set in this macro are
816 used as the default values for the other command line options.
818 @var{level} is the optimization level specified; 2 if @option{-O2} is
819 specified, 1 if @option{-O} is specified, and 0 if neither is specified.
821 @var{size} is nonzero if @option{-Os} is specified and zero otherwise.
823 This macro is run once at program startup and when the optimization
824 options are changed via @code{#pragma GCC optimize} or by using the
825 @code{optimize} attribute.
827 @strong{Do not examine @code{write_symbols} in
828 this macro!} The debugging options are not supposed to alter the
833 This hook is called in response to the user invoking
834 @option{--target-help} on the command line. It gives the target a
835 chance to display extra information on the target specific command
836 line options found in its @file{.opt} file.
839 @defmac CAN_DEBUG_WITHOUT_FP
840 Define this macro if debugging can be performed even without a frame
841 pointer. If this macro is defined, GCC will turn on the
842 @option{-fomit-frame-pointer} option whenever @option{-O} is specified.
845 @defmac SWITCHABLE_TARGET
846 Some targets need to switch between substantially different subtargets
847 during compilation. For example, the MIPS target has one subtarget for
848 the traditional MIPS architecture and another for MIPS16. Source code
849 can switch between these two subarchitectures using the @code{mips16}
850 and @code{nomips16} attributes.
852 Such subtargets can differ in things like the set of available
853 registers, the set of available instructions, the costs of various
854 operations, and so on. GCC caches a lot of this type of information
855 in global variables, and recomputing them for each subtarget takes a
856 significant amount of time. The compiler therefore provides a facility
857 for maintaining several versions of the global variables and quickly
858 switching between them; see @file{target-globals.h} for details.
860 Define this macro to 1 if your target needs this facility. The default
864 @node Per-Function Data
865 @section Defining data structures for per-function information.
866 @cindex per-function data
867 @cindex data structures
869 If the target needs to store information on a per-function basis, GCC
870 provides a macro and a couple of variables to allow this. Note, just
871 using statics to store the information is a bad idea, since GCC supports
872 nested functions, so you can be halfway through encoding one function
873 when another one comes along.
875 GCC defines a data structure called @code{struct function} which
876 contains all of the data specific to an individual function. This
877 structure contains a field called @code{machine} whose type is
878 @code{struct machine_function *}, which can be used by targets to point
879 to their own specific data.
881 If a target needs per-function specific data it should define the type
882 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
883 This macro should be used to initialize the function pointer
884 @code{init_machine_status}. This pointer is explained below.
886 One typical use of per-function, target specific data is to create an
887 RTX to hold the register containing the function's return address. This
888 RTX can then be used to implement the @code{__builtin_return_address}
889 function, for level 0.
891 Note---earlier implementations of GCC used a single data area to hold
892 all of the per-function information. Thus when processing of a nested
893 function began the old per-function data had to be pushed onto a
894 stack, and when the processing was finished, it had to be popped off the
895 stack. GCC used to provide function pointers called
896 @code{save_machine_status} and @code{restore_machine_status} to handle
897 the saving and restoring of the target specific information. Since the
898 single data area approach is no longer used, these pointers are no
901 @defmac INIT_EXPANDERS
902 Macro called to initialize any target specific information. This macro
903 is called once per function, before generation of any RTL has begun.
904 The intention of this macro is to allow the initialization of the
905 function pointer @code{init_machine_status}.
908 @deftypevar {void (*)(struct function *)} init_machine_status
909 If this function pointer is non-@code{NULL} it will be called once per
910 function, before function compilation starts, in order to allow the
911 target to perform any target specific initialization of the
912 @code{struct function} structure. It is intended that this would be
913 used to initialize the @code{machine} of that structure.
915 @code{struct machine_function} structures are expected to be freed by GC@.
916 Generally, any memory that they reference must be allocated by using
917 GC allocation, including the structure itself.
921 @section Storage Layout
922 @cindex storage layout
924 Note that the definitions of the macros in this table which are sizes or
925 alignments measured in bits do not need to be constant. They can be C
926 expressions that refer to static variables, such as the @code{target_flags}.
927 @xref{Run-time Target}.
929 @defmac BITS_BIG_ENDIAN
930 Define this macro to have the value 1 if the most significant bit in a
931 byte has the lowest number; otherwise define it to have the value zero.
932 This means that bit-field instructions count from the most significant
933 bit. If the machine has no bit-field instructions, then this must still
934 be defined, but it doesn't matter which value it is defined to. This
935 macro need not be a constant.
937 This macro does not affect the way structure fields are packed into
938 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
941 @defmac BYTES_BIG_ENDIAN
942 Define this macro to have the value 1 if the most significant byte in a
943 word has the lowest number. This macro need not be a constant.
946 @defmac WORDS_BIG_ENDIAN
947 Define this macro to have the value 1 if, in a multiword object, the
948 most significant word has the lowest number. This applies to both
949 memory locations and registers; GCC fundamentally assumes that the
950 order of words in memory is the same as the order in registers. This
951 macro need not be a constant.
954 @defmac LIBGCC2_WORDS_BIG_ENDIAN
955 Define this macro if @code{WORDS_BIG_ENDIAN} is not constant. This must be a
956 constant value with the same meaning as @code{WORDS_BIG_ENDIAN}, which will be
957 used only when compiling @file{libgcc2.c}. Typically the value will be set
958 based on preprocessor defines.
961 @defmac FLOAT_WORDS_BIG_ENDIAN
962 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
963 @code{TFmode} floating point numbers are stored in memory with the word
964 containing the sign bit at the lowest address; otherwise define it to
965 have the value 0. This macro need not be a constant.
967 You need not define this macro if the ordering is the same as for
971 @defmac BITS_PER_UNIT
972 Define this macro to be the number of bits in an addressable storage
973 unit (byte). If you do not define this macro the default is 8.
976 @defmac BITS_PER_WORD
977 Number of bits in a word. If you do not define this macro, the default
978 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
981 @defmac MAX_BITS_PER_WORD
982 Maximum number of bits in a word. If this is undefined, the default is
983 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
984 largest value that @code{BITS_PER_WORD} can have at run-time.
987 @defmac UNITS_PER_WORD
988 Number of storage units in a word; normally the size of a general-purpose
989 register, a power of two from 1 or 8.
992 @defmac MIN_UNITS_PER_WORD
993 Minimum number of units in a word. If this is undefined, the default is
994 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
995 smallest value that @code{UNITS_PER_WORD} can have at run-time.
998 @defmac UNITS_PER_SIMD_WORD (@var{mode})
999 Number of units in the vectors that the vectorizer can produce for
1000 scalar mode @var{mode}. The default is equal to @code{UNITS_PER_WORD},
1001 because the vectorizer can do some transformations even in absence of
1002 specialized @acronym{SIMD} hardware.
1005 @defmac POINTER_SIZE
1006 Width of a pointer, in bits. You must specify a value no wider than the
1007 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
1008 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
1009 a value the default is @code{BITS_PER_WORD}.
1012 @defmac POINTERS_EXTEND_UNSIGNED
1013 A C expression that determines how pointers should be extended from
1014 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
1015 greater than zero if pointers should be zero-extended, zero if they
1016 should be sign-extended, and negative if some other sort of conversion
1017 is needed. In the last case, the extension is done by the target's
1018 @code{ptr_extend} instruction.
1020 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
1021 and @code{word_mode} are all the same width.
1024 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
1025 A macro to update @var{m} and @var{unsignedp} when an object whose type
1026 is @var{type} and which has the specified mode and signedness is to be
1027 stored in a register. This macro is only called when @var{type} is a
1030 On most RISC machines, which only have operations that operate on a full
1031 register, define this macro to set @var{m} to @code{word_mode} if
1032 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
1033 cases, only integer modes should be widened because wider-precision
1034 floating-point operations are usually more expensive than their narrower
1037 For most machines, the macro definition does not change @var{unsignedp}.
1038 However, some machines, have instructions that preferentially handle
1039 either signed or unsigned quantities of certain modes. For example, on
1040 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
1041 sign-extend the result to 64 bits. On such machines, set
1042 @var{unsignedp} according to which kind of extension is more efficient.
1044 Do not define this macro if it would never modify @var{m}.
1047 @hook TARGET_PROMOTE_FUNCTION_MODE
1048 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
1049 function return values. The target hook should return the new mode
1050 and possibly change @code{*@var{punsignedp}} if the promotion should
1051 change signedness. This function is called only for scalar @emph{or
1054 @var{for_return} allows to distinguish the promotion of arguments and
1055 return values. If it is @code{1}, a return value is being promoted and
1056 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
1057 If it is @code{2}, the returned mode should be that of the register in
1058 which an incoming parameter is copied, or the outgoing result is computed;
1059 then the hook should return the same mode as @code{promote_mode}, though
1060 the signedness may be different.
1062 The default is to not promote arguments and return values. You can
1063 also define the hook to @code{default_promote_function_mode_always_promote}
1064 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
1067 @defmac PARM_BOUNDARY
1068 Normal alignment required for function parameters on the stack, in
1069 bits. All stack parameters receive at least this much alignment
1070 regardless of data type. On most machines, this is the same as the
1074 @defmac STACK_BOUNDARY
1075 Define this macro to the minimum alignment enforced by hardware for the
1076 stack pointer on this machine. The definition is a C expression for the
1077 desired alignment (measured in bits). This value is used as a default
1078 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
1079 this should be the same as @code{PARM_BOUNDARY}.
1082 @defmac PREFERRED_STACK_BOUNDARY
1083 Define this macro if you wish to preserve a certain alignment for the
1084 stack pointer, greater than what the hardware enforces. The definition
1085 is a C expression for the desired alignment (measured in bits). This
1086 macro must evaluate to a value equal to or larger than
1087 @code{STACK_BOUNDARY}.
1090 @defmac INCOMING_STACK_BOUNDARY
1091 Define this macro if the incoming stack boundary may be different
1092 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
1093 to a value equal to or larger than @code{STACK_BOUNDARY}.
1096 @defmac FUNCTION_BOUNDARY
1097 Alignment required for a function entry point, in bits.
1100 @defmac BIGGEST_ALIGNMENT
1101 Biggest alignment that any data type can require on this machine, in
1102 bits. Note that this is not the biggest alignment that is supported,
1103 just the biggest alignment that, when violated, may cause a fault.
1106 @defmac MALLOC_ABI_ALIGNMENT
1107 Alignment, in bits, a C conformant malloc implementation has to
1108 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1111 @defmac ATTRIBUTE_ALIGNED_VALUE
1112 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1113 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1116 @defmac MINIMUM_ATOMIC_ALIGNMENT
1117 If defined, the smallest alignment, in bits, that can be given to an
1118 object that can be referenced in one operation, without disturbing any
1119 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1120 on machines that don't have byte or half-word store operations.
1123 @defmac BIGGEST_FIELD_ALIGNMENT
1124 Biggest alignment that any structure or union field can require on this
1125 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1126 structure and union fields only, unless the field alignment has been set
1127 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1130 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1131 An expression for the alignment of a structure field @var{field} if the
1132 alignment computed in the usual way (including applying of
1133 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1134 alignment) is @var{computed}. It overrides alignment only if the
1135 field alignment has not been set by the
1136 @code{__attribute__ ((aligned (@var{n})))} construct.
1139 @defmac MAX_STACK_ALIGNMENT
1140 Biggest stack alignment guaranteed by the backend. Use this macro
1141 to specify the maximum alignment of a variable on stack.
1143 If not defined, the default value is @code{STACK_BOUNDARY}.
1145 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1146 @c But the fix for PR 32893 indicates that we can only guarantee
1147 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1148 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1151 @defmac MAX_OFILE_ALIGNMENT
1152 Biggest alignment supported by the object file format of this machine.
1153 Use this macro to limit the alignment which can be specified using the
1154 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1155 the default value is @code{BIGGEST_ALIGNMENT}.
1157 On systems that use ELF, the default (in @file{config/elfos.h}) is
1158 the largest supported 32-bit ELF section alignment representable on
1159 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1160 On 32-bit ELF the largest supported section alignment in bits is
1161 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1164 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1165 If defined, a C expression to compute the alignment for a variable in
1166 the static store. @var{type} is the data type, and @var{basic-align} is
1167 the alignment that the object would ordinarily have. The value of this
1168 macro is used instead of that alignment to align the object.
1170 If this macro is not defined, then @var{basic-align} is used.
1173 One use of this macro is to increase alignment of medium-size data to
1174 make it all fit in fewer cache lines. Another is to cause character
1175 arrays to be word-aligned so that @code{strcpy} calls that copy
1176 constants to character arrays can be done inline.
1179 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1180 If defined, a C expression to compute the alignment given to a constant
1181 that is being placed in memory. @var{constant} is the constant and
1182 @var{basic-align} is the alignment that the object would ordinarily
1183 have. The value of this macro is used instead of that alignment to
1186 If this macro is not defined, then @var{basic-align} is used.
1188 The typical use of this macro is to increase alignment for string
1189 constants to be word aligned so that @code{strcpy} calls that copy
1190 constants can be done inline.
1193 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1194 If defined, a C expression to compute the alignment for a variable in
1195 the local store. @var{type} is the data type, and @var{basic-align} is
1196 the alignment that the object would ordinarily have. The value of this
1197 macro is used instead of that alignment to align the object.
1199 If this macro is not defined, then @var{basic-align} is used.
1201 One use of this macro is to increase alignment of medium-size data to
1202 make it all fit in fewer cache lines.
1205 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1206 If defined, a C expression to compute the alignment for stack slot.
1207 @var{type} is the data type, @var{mode} is the widest mode available,
1208 and @var{basic-align} is the alignment that the slot would ordinarily
1209 have. The value of this macro is used instead of that alignment to
1212 If this macro is not defined, then @var{basic-align} is used when
1213 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1216 This macro is to set alignment of stack slot to the maximum alignment
1217 of all possible modes which the slot may have.
1220 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1221 If defined, a C expression to compute the alignment for a local
1222 variable @var{decl}.
1224 If this macro is not defined, then
1225 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1228 One use of this macro is to increase alignment of medium-size data to
1229 make it all fit in fewer cache lines.
1232 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1233 If defined, a C expression to compute the minimum required alignment
1234 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1235 @var{mode}, assuming normal alignment @var{align}.
1237 If this macro is not defined, then @var{align} will be used.
1240 @defmac EMPTY_FIELD_BOUNDARY
1241 Alignment in bits to be given to a structure bit-field that follows an
1242 empty field such as @code{int : 0;}.
1244 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1247 @defmac STRUCTURE_SIZE_BOUNDARY
1248 Number of bits which any structure or union's size must be a multiple of.
1249 Each structure or union's size is rounded up to a multiple of this.
1251 If you do not define this macro, the default is the same as
1252 @code{BITS_PER_UNIT}.
1255 @defmac STRICT_ALIGNMENT
1256 Define this macro to be the value 1 if instructions will fail to work
1257 if given data not on the nominal alignment. If instructions will merely
1258 go slower in that case, define this macro as 0.
1261 @defmac PCC_BITFIELD_TYPE_MATTERS
1262 Define this if you wish to imitate the way many other C compilers handle
1263 alignment of bit-fields and the structures that contain them.
1265 The behavior is that the type written for a named bit-field (@code{int},
1266 @code{short}, or other integer type) imposes an alignment for the entire
1267 structure, as if the structure really did contain an ordinary field of
1268 that type. In addition, the bit-field is placed within the structure so
1269 that it would fit within such a field, not crossing a boundary for it.
1271 Thus, on most machines, a named bit-field whose type is written as
1272 @code{int} would not cross a four-byte boundary, and would force
1273 four-byte alignment for the whole structure. (The alignment used may
1274 not be four bytes; it is controlled by the other alignment parameters.)
1276 An unnamed bit-field will not affect the alignment of the containing
1279 If the macro is defined, its definition should be a C expression;
1280 a nonzero value for the expression enables this behavior.
1282 Note that if this macro is not defined, or its value is zero, some
1283 bit-fields may cross more than one alignment boundary. The compiler can
1284 support such references if there are @samp{insv}, @samp{extv}, and
1285 @samp{extzv} insns that can directly reference memory.
1287 The other known way of making bit-fields work is to define
1288 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1289 Then every structure can be accessed with fullwords.
1291 Unless the machine has bit-field instructions or you define
1292 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1293 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1295 If your aim is to make GCC use the same conventions for laying out
1296 bit-fields as are used by another compiler, here is how to investigate
1297 what the other compiler does. Compile and run this program:
1316 printf ("Size of foo1 is %d\n",
1317 sizeof (struct foo1));
1318 printf ("Size of foo2 is %d\n",
1319 sizeof (struct foo2));
1324 If this prints 2 and 5, then the compiler's behavior is what you would
1325 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1328 @defmac BITFIELD_NBYTES_LIMITED
1329 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1330 to aligning a bit-field within the structure.
1333 @hook TARGET_ALIGN_ANON_BITFIELD
1334 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1335 whether unnamed bitfields affect the alignment of the containing
1336 structure. The hook should return true if the structure should inherit
1337 the alignment requirements of an unnamed bitfield's type.
1340 @hook TARGET_NARROW_VOLATILE_BITFIELD
1341 This target hook should return @code{true} if accesses to volatile bitfields
1342 should use the narrowest mode possible. It should return @code{false} if
1343 these accesses should use the bitfield container type.
1345 The default is @code{!TARGET_STRICT_ALIGN}.
1348 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1349 Return 1 if a structure or array containing @var{field} should be accessed using
1352 If @var{field} is the only field in the structure, @var{mode} is its
1353 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1354 case where structures of one field would require the structure's mode to
1355 retain the field's mode.
1357 Normally, this is not needed.
1360 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1361 Define this macro as an expression for the alignment of a type (given
1362 by @var{type} as a tree node) if the alignment computed in the usual
1363 way is @var{computed} and the alignment explicitly specified was
1366 The default is to use @var{specified} if it is larger; otherwise, use
1367 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1370 @defmac MAX_FIXED_MODE_SIZE
1371 An integer expression for the size in bits of the largest integer
1372 machine mode that should actually be used. All integer machine modes of
1373 this size or smaller can be used for structures and unions with the
1374 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1375 (DImode)} is assumed.
1378 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1379 If defined, an expression of type @code{enum machine_mode} that
1380 specifies the mode of the save area operand of a
1381 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1382 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1383 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1384 having its mode specified.
1386 You need not define this macro if it always returns @code{Pmode}. You
1387 would most commonly define this macro if the
1388 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1392 @defmac STACK_SIZE_MODE
1393 If defined, an expression of type @code{enum machine_mode} that
1394 specifies the mode of the size increment operand of an
1395 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1397 You need not define this macro if it always returns @code{word_mode}.
1398 You would most commonly define this macro if the @code{allocate_stack}
1399 pattern needs to support both a 32- and a 64-bit mode.
1402 @hook TARGET_LIBGCC_CMP_RETURN_MODE
1403 This target hook should return the mode to be used for the return value
1404 of compare instructions expanded to libgcc calls. If not defined
1405 @code{word_mode} is returned which is the right choice for a majority of
1409 @hook TARGET_LIBGCC_SHIFT_COUNT_MODE
1410 This target hook should return the mode to be used for the shift count operand
1411 of shift instructions expanded to libgcc calls. If not defined
1412 @code{word_mode} is returned which is the right choice for a majority of
1416 @hook TARGET_UNWIND_WORD_MODE
1417 Return machine mode to be used for @code{_Unwind_Word} type.
1418 The default is to use @code{word_mode}.
1421 @defmac ROUND_TOWARDS_ZERO
1422 If defined, this macro should be true if the prevailing rounding
1423 mode is towards zero.
1425 Defining this macro only affects the way @file{libgcc.a} emulates
1426 floating-point arithmetic.
1428 Not defining this macro is equivalent to returning zero.
1431 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1432 This macro should return true if floats with @var{size}
1433 bits do not have a NaN or infinity representation, but use the largest
1434 exponent for normal numbers instead.
1436 Defining this macro only affects the way @file{libgcc.a} emulates
1437 floating-point arithmetic.
1439 The default definition of this macro returns false for all sizes.
1442 @hook TARGET_MS_BITFIELD_LAYOUT_P
1443 This target hook returns @code{true} if bit-fields in the given
1444 @var{record_type} are to be laid out following the rules of Microsoft
1445 Visual C/C++, namely: (i) a bit-field won't share the same storage
1446 unit with the previous bit-field if their underlying types have
1447 different sizes, and the bit-field will be aligned to the highest
1448 alignment of the underlying types of itself and of the previous
1449 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1450 the whole enclosing structure, even if it is unnamed; except that
1451 (iii) a zero-sized bit-field will be disregarded unless it follows
1452 another bit-field of nonzero size. If this hook returns @code{true},
1453 other macros that control bit-field layout are ignored.
1455 When a bit-field is inserted into a packed record, the whole size
1456 of the underlying type is used by one or more same-size adjacent
1457 bit-fields (that is, if its long:3, 32 bits is used in the record,
1458 and any additional adjacent long bit-fields are packed into the same
1459 chunk of 32 bits. However, if the size changes, a new field of that
1460 size is allocated). In an unpacked record, this is the same as using
1461 alignment, but not equivalent when packing.
1463 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1464 the latter will take precedence. If @samp{__attribute__((packed))} is
1465 used on a single field when MS bit-fields are in use, it will take
1466 precedence for that field, but the alignment of the rest of the structure
1467 may affect its placement.
1470 @hook TARGET_DECIMAL_FLOAT_SUPPORTED_P
1471 Returns true if the target supports decimal floating point.
1474 @hook TARGET_FIXED_POINT_SUPPORTED_P
1475 Returns true if the target supports fixed-point arithmetic.
1478 @hook TARGET_EXPAND_TO_RTL_HOOK
1479 This hook is called just before expansion into rtl, allowing the target
1480 to perform additional initializations or analysis before the expansion.
1481 For example, the rs6000 port uses it to allocate a scratch stack slot
1482 for use in copying SDmode values between memory and floating point
1483 registers whenever the function being expanded has any SDmode
1487 @hook TARGET_INSTANTIATE_DECLS
1488 This hook allows the backend to perform additional instantiations on rtl
1489 that are not actually in any insns yet, but will be later.
1492 @hook TARGET_MANGLE_TYPE
1493 If your target defines any fundamental types, or any types your target
1494 uses should be mangled differently from the default, define this hook
1495 to return the appropriate encoding for these types as part of a C++
1496 mangled name. The @var{type} argument is the tree structure representing
1497 the type to be mangled. The hook may be applied to trees which are
1498 not target-specific fundamental types; it should return @code{NULL}
1499 for all such types, as well as arguments it does not recognize. If the
1500 return value is not @code{NULL}, it must point to a statically-allocated
1503 Target-specific fundamental types might be new fundamental types or
1504 qualified versions of ordinary fundamental types. Encode new
1505 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1506 is the name used for the type in source code, and @var{n} is the
1507 length of @var{name} in decimal. Encode qualified versions of
1508 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1509 @var{name} is the name used for the type qualifier in source code,
1510 @var{n} is the length of @var{name} as above, and @var{code} is the
1511 code used to represent the unqualified version of this type. (See
1512 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1513 codes.) In both cases the spaces are for clarity; do not include any
1514 spaces in your string.
1516 This hook is applied to types prior to typedef resolution. If the mangled
1517 name for a particular type depends only on that type's main variant, you
1518 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1521 The default version of this hook always returns @code{NULL}, which is
1522 appropriate for a target that does not define any new fundamental
1527 @section Layout of Source Language Data Types
1529 These macros define the sizes and other characteristics of the standard
1530 basic data types used in programs being compiled. Unlike the macros in
1531 the previous section, these apply to specific features of C and related
1532 languages, rather than to fundamental aspects of storage layout.
1534 @defmac INT_TYPE_SIZE
1535 A C expression for the size in bits of the type @code{int} on the
1536 target machine. If you don't define this, the default is one word.
1539 @defmac SHORT_TYPE_SIZE
1540 A C expression for the size in bits of the type @code{short} on the
1541 target machine. If you don't define this, the default is half a word.
1542 (If this would be less than one storage unit, it is rounded up to one
1546 @defmac LONG_TYPE_SIZE
1547 A C expression for the size in bits of the type @code{long} on the
1548 target machine. If you don't define this, the default is one word.
1551 @defmac ADA_LONG_TYPE_SIZE
1552 On some machines, the size used for the Ada equivalent of the type
1553 @code{long} by a native Ada compiler differs from that used by C@. In
1554 that situation, define this macro to be a C expression to be used for
1555 the size of that type. If you don't define this, the default is the
1556 value of @code{LONG_TYPE_SIZE}.
1559 @defmac LONG_LONG_TYPE_SIZE
1560 A C expression for the size in bits of the type @code{long long} on the
1561 target machine. If you don't define this, the default is two
1562 words. If you want to support GNU Ada on your machine, the value of this
1563 macro must be at least 64.
1566 @defmac CHAR_TYPE_SIZE
1567 A C expression for the size in bits of the type @code{char} on the
1568 target machine. If you don't define this, the default is
1569 @code{BITS_PER_UNIT}.
1572 @defmac BOOL_TYPE_SIZE
1573 A C expression for the size in bits of the C++ type @code{bool} and
1574 C99 type @code{_Bool} on the target machine. If you don't define
1575 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1578 @defmac FLOAT_TYPE_SIZE
1579 A C expression for the size in bits of the type @code{float} on the
1580 target machine. If you don't define this, the default is one word.
1583 @defmac DOUBLE_TYPE_SIZE
1584 A C expression for the size in bits of the type @code{double} on the
1585 target machine. If you don't define this, the default is two
1589 @defmac LONG_DOUBLE_TYPE_SIZE
1590 A C expression for the size in bits of the type @code{long double} on
1591 the target machine. If you don't define this, the default is two
1595 @defmac SHORT_FRACT_TYPE_SIZE
1596 A C expression for the size in bits of the type @code{short _Fract} on
1597 the target machine. If you don't define this, the default is
1598 @code{BITS_PER_UNIT}.
1601 @defmac FRACT_TYPE_SIZE
1602 A C expression for the size in bits of the type @code{_Fract} on
1603 the target machine. If you don't define this, the default is
1604 @code{BITS_PER_UNIT * 2}.
1607 @defmac LONG_FRACT_TYPE_SIZE
1608 A C expression for the size in bits of the type @code{long _Fract} on
1609 the target machine. If you don't define this, the default is
1610 @code{BITS_PER_UNIT * 4}.
1613 @defmac LONG_LONG_FRACT_TYPE_SIZE
1614 A C expression for the size in bits of the type @code{long long _Fract} on
1615 the target machine. If you don't define this, the default is
1616 @code{BITS_PER_UNIT * 8}.
1619 @defmac SHORT_ACCUM_TYPE_SIZE
1620 A C expression for the size in bits of the type @code{short _Accum} on
1621 the target machine. If you don't define this, the default is
1622 @code{BITS_PER_UNIT * 2}.
1625 @defmac ACCUM_TYPE_SIZE
1626 A C expression for the size in bits of the type @code{_Accum} on
1627 the target machine. If you don't define this, the default is
1628 @code{BITS_PER_UNIT * 4}.
1631 @defmac LONG_ACCUM_TYPE_SIZE
1632 A C expression for the size in bits of the type @code{long _Accum} on
1633 the target machine. If you don't define this, the default is
1634 @code{BITS_PER_UNIT * 8}.
1637 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1638 A C expression for the size in bits of the type @code{long long _Accum} on
1639 the target machine. If you don't define this, the default is
1640 @code{BITS_PER_UNIT * 16}.
1643 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1644 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1645 if you want routines in @file{libgcc2.a} for a size other than
1646 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1647 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1650 @defmac LIBGCC2_HAS_DF_MODE
1651 Define this macro if neither @code{LIBGCC2_DOUBLE_TYPE_SIZE} nor
1652 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1653 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1654 anyway. If you don't define this and either @code{LIBGCC2_DOUBLE_TYPE_SIZE}
1655 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1659 @defmac LIBGCC2_HAS_XF_MODE
1660 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1661 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1662 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1663 is 80 then the default is 1, otherwise it is 0.
1666 @defmac LIBGCC2_HAS_TF_MODE
1667 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1668 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1669 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1670 is 128 then the default is 1, otherwise it is 0.
1677 Define these macros to be the size in bits of the mantissa of
1678 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1679 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1680 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1681 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1682 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1683 @code{LIBGCC2_DOUBLE_TYPE_SIZE} or
1684 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1687 @defmac TARGET_FLT_EVAL_METHOD
1688 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1689 assuming, if applicable, that the floating-point control word is in its
1690 default state. If you do not define this macro the value of
1691 @code{FLT_EVAL_METHOD} will be zero.
1694 @defmac WIDEST_HARDWARE_FP_SIZE
1695 A C expression for the size in bits of the widest floating-point format
1696 supported by the hardware. If you define this macro, you must specify a
1697 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1698 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1702 @defmac DEFAULT_SIGNED_CHAR
1703 An expression whose value is 1 or 0, according to whether the type
1704 @code{char} should be signed or unsigned by default. The user can
1705 always override this default with the options @option{-fsigned-char}
1706 and @option{-funsigned-char}.
1709 @hook TARGET_DEFAULT_SHORT_ENUMS
1710 This target hook should return true if the compiler should give an
1711 @code{enum} type only as many bytes as it takes to represent the range
1712 of possible values of that type. It should return false if all
1713 @code{enum} types should be allocated like @code{int}.
1715 The default is to return false.
1719 A C expression for a string describing the name of the data type to use
1720 for size values. The typedef name @code{size_t} is defined using the
1721 contents of the string.
1723 The string can contain more than one keyword. If so, separate them with
1724 spaces, and write first any length keyword, then @code{unsigned} if
1725 appropriate, and finally @code{int}. The string must exactly match one
1726 of the data type names defined in the function
1727 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1728 omit @code{int} or change the order---that would cause the compiler to
1731 If you don't define this macro, the default is @code{"long unsigned
1735 @defmac PTRDIFF_TYPE
1736 A C expression for a string describing the name of the data type to use
1737 for the result of subtracting two pointers. The typedef name
1738 @code{ptrdiff_t} is defined using the contents of the string. See
1739 @code{SIZE_TYPE} above for more information.
1741 If you don't define this macro, the default is @code{"long int"}.
1745 A C expression for a string describing the name of the data type to use
1746 for wide characters. The typedef name @code{wchar_t} is defined using
1747 the contents of the string. See @code{SIZE_TYPE} above for more
1750 If you don't define this macro, the default is @code{"int"}.
1753 @defmac WCHAR_TYPE_SIZE
1754 A C expression for the size in bits of the data type for wide
1755 characters. This is used in @code{cpp}, which cannot make use of
1760 A C expression for a string describing the name of the data type to
1761 use for wide characters passed to @code{printf} and returned from
1762 @code{getwc}. The typedef name @code{wint_t} is defined using the
1763 contents of the string. See @code{SIZE_TYPE} above for more
1766 If you don't define this macro, the default is @code{"unsigned int"}.
1770 A C expression for a string describing the name of the data type that
1771 can represent any value of any standard or extended signed integer type.
1772 The typedef name @code{intmax_t} is defined using the contents of the
1773 string. See @code{SIZE_TYPE} above for more information.
1775 If you don't define this macro, the default is the first of
1776 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1777 much precision as @code{long long int}.
1780 @defmac UINTMAX_TYPE
1781 A C expression for a string describing the name of the data type that
1782 can represent any value of any standard or extended unsigned integer
1783 type. The typedef name @code{uintmax_t} is defined using the contents
1784 of the string. See @code{SIZE_TYPE} above for more information.
1786 If you don't define this macro, the default is the first of
1787 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1788 unsigned int"} that has as much precision as @code{long long unsigned
1792 @defmac SIG_ATOMIC_TYPE
1798 @defmacx UINT16_TYPE
1799 @defmacx UINT32_TYPE
1800 @defmacx UINT64_TYPE
1801 @defmacx INT_LEAST8_TYPE
1802 @defmacx INT_LEAST16_TYPE
1803 @defmacx INT_LEAST32_TYPE
1804 @defmacx INT_LEAST64_TYPE
1805 @defmacx UINT_LEAST8_TYPE
1806 @defmacx UINT_LEAST16_TYPE
1807 @defmacx UINT_LEAST32_TYPE
1808 @defmacx UINT_LEAST64_TYPE
1809 @defmacx INT_FAST8_TYPE
1810 @defmacx INT_FAST16_TYPE
1811 @defmacx INT_FAST32_TYPE
1812 @defmacx INT_FAST64_TYPE
1813 @defmacx UINT_FAST8_TYPE
1814 @defmacx UINT_FAST16_TYPE
1815 @defmacx UINT_FAST32_TYPE
1816 @defmacx UINT_FAST64_TYPE
1817 @defmacx INTPTR_TYPE
1818 @defmacx UINTPTR_TYPE
1819 C expressions for the standard types @code{sig_atomic_t},
1820 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1821 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1822 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1823 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1824 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1825 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1826 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1827 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1828 @code{SIZE_TYPE} above for more information.
1830 If any of these macros evaluates to a null pointer, the corresponding
1831 type is not supported; if GCC is configured to provide
1832 @code{<stdint.h>} in such a case, the header provided may not conform
1833 to C99, depending on the type in question. The defaults for all of
1834 these macros are null pointers.
1837 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1838 The C++ compiler represents a pointer-to-member-function with a struct
1845 ptrdiff_t vtable_index;
1852 The C++ compiler must use one bit to indicate whether the function that
1853 will be called through a pointer-to-member-function is virtual.
1854 Normally, we assume that the low-order bit of a function pointer must
1855 always be zero. Then, by ensuring that the vtable_index is odd, we can
1856 distinguish which variant of the union is in use. But, on some
1857 platforms function pointers can be odd, and so this doesn't work. In
1858 that case, we use the low-order bit of the @code{delta} field, and shift
1859 the remainder of the @code{delta} field to the left.
1861 GCC will automatically make the right selection about where to store
1862 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1863 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1864 set such that functions always start at even addresses, but the lowest
1865 bit of pointers to functions indicate whether the function at that
1866 address is in ARM or Thumb mode. If this is the case of your
1867 architecture, you should define this macro to
1868 @code{ptrmemfunc_vbit_in_delta}.
1870 In general, you should not have to define this macro. On architectures
1871 in which function addresses are always even, according to
1872 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1873 @code{ptrmemfunc_vbit_in_pfn}.
1876 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1877 Normally, the C++ compiler uses function pointers in vtables. This
1878 macro allows the target to change to use ``function descriptors''
1879 instead. Function descriptors are found on targets for whom a
1880 function pointer is actually a small data structure. Normally the
1881 data structure consists of the actual code address plus a data
1882 pointer to which the function's data is relative.
1884 If vtables are used, the value of this macro should be the number
1885 of words that the function descriptor occupies.
1888 @defmac TARGET_VTABLE_ENTRY_ALIGN
1889 By default, the vtable entries are void pointers, the so the alignment
1890 is the same as pointer alignment. The value of this macro specifies
1891 the alignment of the vtable entry in bits. It should be defined only
1892 when special alignment is necessary. */
1895 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1896 There are a few non-descriptor entries in the vtable at offsets below
1897 zero. If these entries must be padded (say, to preserve the alignment
1898 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1899 of words in each data entry.
1903 @section Register Usage
1904 @cindex register usage
1906 This section explains how to describe what registers the target machine
1907 has, and how (in general) they can be used.
1909 The description of which registers a specific instruction can use is
1910 done with register classes; see @ref{Register Classes}. For information
1911 on using registers to access a stack frame, see @ref{Frame Registers}.
1912 For passing values in registers, see @ref{Register Arguments}.
1913 For returning values in registers, see @ref{Scalar Return}.
1916 * Register Basics:: Number and kinds of registers.
1917 * Allocation Order:: Order in which registers are allocated.
1918 * Values in Registers:: What kinds of values each reg can hold.
1919 * Leaf Functions:: Renumbering registers for leaf functions.
1920 * Stack Registers:: Handling a register stack such as 80387.
1923 @node Register Basics
1924 @subsection Basic Characteristics of Registers
1926 @c prevent bad page break with this line
1927 Registers have various characteristics.
1929 @defmac FIRST_PSEUDO_REGISTER
1930 Number of hardware registers known to the compiler. They receive
1931 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1932 pseudo register's number really is assigned the number
1933 @code{FIRST_PSEUDO_REGISTER}.
1936 @defmac FIXED_REGISTERS
1937 @cindex fixed register
1938 An initializer that says which registers are used for fixed purposes
1939 all throughout the compiled code and are therefore not available for
1940 general allocation. These would include the stack pointer, the frame
1941 pointer (except on machines where that can be used as a general
1942 register when no frame pointer is needed), the program counter on
1943 machines where that is considered one of the addressable registers,
1944 and any other numbered register with a standard use.
1946 This information is expressed as a sequence of numbers, separated by
1947 commas and surrounded by braces. The @var{n}th number is 1 if
1948 register @var{n} is fixed, 0 otherwise.
1950 The table initialized from this macro, and the table initialized by
1951 the following one, may be overridden at run time either automatically,
1952 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1953 the user with the command options @option{-ffixed-@var{reg}},
1954 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1957 @defmac CALL_USED_REGISTERS
1958 @cindex call-used register
1959 @cindex call-clobbered register
1960 @cindex call-saved register
1961 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1962 clobbered (in general) by function calls as well as for fixed
1963 registers. This macro therefore identifies the registers that are not
1964 available for general allocation of values that must live across
1967 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1968 automatically saves it on function entry and restores it on function
1969 exit, if the register is used within the function.
1972 @defmac CALL_REALLY_USED_REGISTERS
1973 @cindex call-used register
1974 @cindex call-clobbered register
1975 @cindex call-saved register
1976 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1977 that the entire set of @code{FIXED_REGISTERS} be included.
1978 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1979 This macro is optional. If not specified, it defaults to the value
1980 of @code{CALL_USED_REGISTERS}.
1983 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1984 @cindex call-used register
1985 @cindex call-clobbered register
1986 @cindex call-saved register
1987 A C expression that is nonzero if it is not permissible to store a
1988 value of mode @var{mode} in hard register number @var{regno} across a
1989 call without some part of it being clobbered. For most machines this
1990 macro need not be defined. It is only required for machines that do not
1991 preserve the entire contents of a register across a call.
1995 @findex call_used_regs
1998 @findex reg_class_contents
1999 @defmac CONDITIONAL_REGISTER_USAGE
2000 Zero or more C statements that may conditionally modify five variables
2001 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
2002 @code{reg_names}, and @code{reg_class_contents}, to take into account
2003 any dependence of these register sets on target flags. The first three
2004 of these are of type @code{char []} (interpreted as Boolean vectors).
2005 @code{global_regs} is a @code{const char *[]}, and
2006 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
2007 called, @code{fixed_regs}, @code{call_used_regs},
2008 @code{reg_class_contents}, and @code{reg_names} have been initialized
2009 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
2010 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
2011 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
2012 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
2013 command options have been applied.
2015 You need not define this macro if it has no work to do.
2017 @cindex disabling certain registers
2018 @cindex controlling register usage
2019 If the usage of an entire class of registers depends on the target
2020 flags, you may indicate this to GCC by using this macro to modify
2021 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
2022 registers in the classes which should not be used by GCC@. Also define
2023 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
2024 to return @code{NO_REGS} if it
2025 is called with a letter for a class that shouldn't be used.
2027 (However, if this class is not included in @code{GENERAL_REGS} and all
2028 of the insn patterns whose constraints permit this class are
2029 controlled by target switches, then GCC will automatically avoid using
2030 these registers when the target switches are opposed to them.)
2033 @defmac INCOMING_REGNO (@var{out})
2034 Define this macro if the target machine has register windows. This C
2035 expression returns the register number as seen by the called function
2036 corresponding to the register number @var{out} as seen by the calling
2037 function. Return @var{out} if register number @var{out} is not an
2041 @defmac OUTGOING_REGNO (@var{in})
2042 Define this macro if the target machine has register windows. This C
2043 expression returns the register number as seen by the calling function
2044 corresponding to the register number @var{in} as seen by the called
2045 function. Return @var{in} if register number @var{in} is not an inbound
2049 @defmac LOCAL_REGNO (@var{regno})
2050 Define this macro if the target machine has register windows. This C
2051 expression returns true if the register is call-saved but is in the
2052 register window. Unlike most call-saved registers, such registers
2053 need not be explicitly restored on function exit or during non-local
2058 If the program counter has a register number, define this as that
2059 register number. Otherwise, do not define it.
2062 @node Allocation Order
2063 @subsection Order of Allocation of Registers
2064 @cindex order of register allocation
2065 @cindex register allocation order
2067 @c prevent bad page break with this line
2068 Registers are allocated in order.
2070 @defmac REG_ALLOC_ORDER
2071 If defined, an initializer for a vector of integers, containing the
2072 numbers of hard registers in the order in which GCC should prefer
2073 to use them (from most preferred to least).
2075 If this macro is not defined, registers are used lowest numbered first
2076 (all else being equal).
2078 One use of this macro is on machines where the highest numbered
2079 registers must always be saved and the save-multiple-registers
2080 instruction supports only sequences of consecutive registers. On such
2081 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
2082 the highest numbered allocable register first.
2085 @defmac ADJUST_REG_ALLOC_ORDER
2086 A C statement (sans semicolon) to choose the order in which to allocate
2087 hard registers for pseudo-registers local to a basic block.
2089 Store the desired register order in the array @code{reg_alloc_order}.
2090 Element 0 should be the register to allocate first; element 1, the next
2091 register; and so on.
2093 The macro body should not assume anything about the contents of
2094 @code{reg_alloc_order} before execution of the macro.
2096 On most machines, it is not necessary to define this macro.
2099 @defmac HONOR_REG_ALLOC_ORDER
2100 Normally, IRA tries to estimate the costs for saving a register in the
2101 prologue and restoring it in the epilogue. This discourages it from
2102 using call-saved registers. If a machine wants to ensure that IRA
2103 allocates registers in the order given by REG_ALLOC_ORDER even if some
2104 call-saved registers appear earlier than call-used ones, this macro
2108 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
2109 In some case register allocation order is not enough for the
2110 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
2111 If this macro is defined, it should return a floating point value
2112 based on @var{regno}. The cost of using @var{regno} for a pseudo will
2113 be increased by approximately the pseudo's usage frequency times the
2114 value returned by this macro. Not defining this macro is equivalent
2115 to having it always return @code{0.0}.
2117 On most machines, it is not necessary to define this macro.
2120 @node Values in Registers
2121 @subsection How Values Fit in Registers
2123 This section discusses the macros that describe which kinds of values
2124 (specifically, which machine modes) each register can hold, and how many
2125 consecutive registers are needed for a given mode.
2127 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2128 A C expression for the number of consecutive hard registers, starting
2129 at register number @var{regno}, required to hold a value of mode
2130 @var{mode}. This macro must never return zero, even if a register
2131 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2132 and/or CANNOT_CHANGE_MODE_CLASS instead.
2134 On a machine where all registers are exactly one word, a suitable
2135 definition of this macro is
2138 #define HARD_REGNO_NREGS(REGNO, MODE) \
2139 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2144 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2145 A C expression that is nonzero if a value of mode @var{mode}, stored
2146 in memory, ends with padding that causes it to take up more space than
2147 in registers starting at register number @var{regno} (as determined by
2148 multiplying GCC's notion of the size of the register when containing
2149 this mode by the number of registers returned by
2150 @code{HARD_REGNO_NREGS}). By default this is zero.
2152 For example, if a floating-point value is stored in three 32-bit
2153 registers but takes up 128 bits in memory, then this would be
2156 This macros only needs to be defined if there are cases where
2157 @code{subreg_get_info}
2158 would otherwise wrongly determine that a @code{subreg} can be
2159 represented by an offset to the register number, when in fact such a
2160 @code{subreg} would contain some of the padding not stored in
2161 registers and so not be representable.
2164 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2165 For values of @var{regno} and @var{mode} for which
2166 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2167 returning the greater number of registers required to hold the value
2168 including any padding. In the example above, the value would be four.
2171 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2172 Define this macro if the natural size of registers that hold values
2173 of mode @var{mode} is not the word size. It is a C expression that
2174 should give the natural size in bytes for the specified mode. It is
2175 used by the register allocator to try to optimize its results. This
2176 happens for example on SPARC 64-bit where the natural size of
2177 floating-point registers is still 32-bit.
2180 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2181 A C expression that is nonzero if it is permissible to store a value
2182 of mode @var{mode} in hard register number @var{regno} (or in several
2183 registers starting with that one). For a machine where all registers
2184 are equivalent, a suitable definition is
2187 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2190 You need not include code to check for the numbers of fixed registers,
2191 because the allocation mechanism considers them to be always occupied.
2193 @cindex register pairs
2194 On some machines, double-precision values must be kept in even/odd
2195 register pairs. You can implement that by defining this macro to reject
2196 odd register numbers for such modes.
2198 The minimum requirement for a mode to be OK in a register is that the
2199 @samp{mov@var{mode}} instruction pattern support moves between the
2200 register and other hard register in the same class and that moving a
2201 value into the register and back out not alter it.
2203 Since the same instruction used to move @code{word_mode} will work for
2204 all narrower integer modes, it is not necessary on any machine for
2205 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2206 you define patterns @samp{movhi}, etc., to take advantage of this. This
2207 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2208 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2211 Many machines have special registers for floating point arithmetic.
2212 Often people assume that floating point machine modes are allowed only
2213 in floating point registers. This is not true. Any registers that
2214 can hold integers can safely @emph{hold} a floating point machine
2215 mode, whether or not floating arithmetic can be done on it in those
2216 registers. Integer move instructions can be used to move the values.
2218 On some machines, though, the converse is true: fixed-point machine
2219 modes may not go in floating registers. This is true if the floating
2220 registers normalize any value stored in them, because storing a
2221 non-floating value there would garble it. In this case,
2222 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2223 floating registers. But if the floating registers do not automatically
2224 normalize, if you can store any bit pattern in one and retrieve it
2225 unchanged without a trap, then any machine mode may go in a floating
2226 register, so you can define this macro to say so.
2228 The primary significance of special floating registers is rather that
2229 they are the registers acceptable in floating point arithmetic
2230 instructions. However, this is of no concern to
2231 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2232 constraints for those instructions.
2234 On some machines, the floating registers are especially slow to access,
2235 so that it is better to store a value in a stack frame than in such a
2236 register if floating point arithmetic is not being done. As long as the
2237 floating registers are not in class @code{GENERAL_REGS}, they will not
2238 be used unless some pattern's constraint asks for one.
2241 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2242 A C expression that is nonzero if it is OK to rename a hard register
2243 @var{from} to another hard register @var{to}.
2245 One common use of this macro is to prevent renaming of a register to
2246 another register that is not saved by a prologue in an interrupt
2249 The default is always nonzero.
2252 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2253 A C expression that is nonzero if a value of mode
2254 @var{mode1} is accessible in mode @var{mode2} without copying.
2256 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2257 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2258 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2259 should be nonzero. If they differ for any @var{r}, you should define
2260 this macro to return zero unless some other mechanism ensures the
2261 accessibility of the value in a narrower mode.
2263 You should define this macro to return nonzero in as many cases as
2264 possible since doing so will allow GCC to perform better register
2268 @hook TARGET_HARD_REGNO_SCRATCH_OK
2269 This target hook should return @code{true} if it is OK to use a hard register
2270 @var{regno} as scratch reg in peephole2.
2272 One common use of this macro is to prevent using of a register that
2273 is not saved by a prologue in an interrupt handler.
2275 The default version of this hook always returns @code{true}.
2278 @defmac AVOID_CCMODE_COPIES
2279 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2280 registers. You should only define this macro if support for copying to/from
2281 @code{CCmode} is incomplete.
2284 @node Leaf Functions
2285 @subsection Handling Leaf Functions
2287 @cindex leaf functions
2288 @cindex functions, leaf
2289 On some machines, a leaf function (i.e., one which makes no calls) can run
2290 more efficiently if it does not make its own register window. Often this
2291 means it is required to receive its arguments in the registers where they
2292 are passed by the caller, instead of the registers where they would
2295 The special treatment for leaf functions generally applies only when
2296 other conditions are met; for example, often they may use only those
2297 registers for its own variables and temporaries. We use the term ``leaf
2298 function'' to mean a function that is suitable for this special
2299 handling, so that functions with no calls are not necessarily ``leaf
2302 GCC assigns register numbers before it knows whether the function is
2303 suitable for leaf function treatment. So it needs to renumber the
2304 registers in order to output a leaf function. The following macros
2307 @defmac LEAF_REGISTERS
2308 Name of a char vector, indexed by hard register number, which
2309 contains 1 for a register that is allowable in a candidate for leaf
2312 If leaf function treatment involves renumbering the registers, then the
2313 registers marked here should be the ones before renumbering---those that
2314 GCC would ordinarily allocate. The registers which will actually be
2315 used in the assembler code, after renumbering, should not be marked with 1
2318 Define this macro only if the target machine offers a way to optimize
2319 the treatment of leaf functions.
2322 @defmac LEAF_REG_REMAP (@var{regno})
2323 A C expression whose value is the register number to which @var{regno}
2324 should be renumbered, when a function is treated as a leaf function.
2326 If @var{regno} is a register number which should not appear in a leaf
2327 function before renumbering, then the expression should yield @minus{}1, which
2328 will cause the compiler to abort.
2330 Define this macro only if the target machine offers a way to optimize the
2331 treatment of leaf functions, and registers need to be renumbered to do
2335 @findex current_function_is_leaf
2336 @findex current_function_uses_only_leaf_regs
2337 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2338 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2339 specially. They can test the C variable @code{current_function_is_leaf}
2340 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2341 set prior to local register allocation and is valid for the remaining
2342 compiler passes. They can also test the C variable
2343 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2344 functions which only use leaf registers.
2345 @code{current_function_uses_only_leaf_regs} is valid after all passes
2346 that modify the instructions have been run and is only useful if
2347 @code{LEAF_REGISTERS} is defined.
2348 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2349 @c of the next paragraph?! --mew 2feb93
2351 @node Stack Registers
2352 @subsection Registers That Form a Stack
2354 There are special features to handle computers where some of the
2355 ``registers'' form a stack. Stack registers are normally written by
2356 pushing onto the stack, and are numbered relative to the top of the
2359 Currently, GCC can only handle one group of stack-like registers, and
2360 they must be consecutively numbered. Furthermore, the existing
2361 support for stack-like registers is specific to the 80387 floating
2362 point coprocessor. If you have a new architecture that uses
2363 stack-like registers, you will need to do substantial work on
2364 @file{reg-stack.c} and write your machine description to cooperate
2365 with it, as well as defining these macros.
2368 Define this if the machine has any stack-like registers.
2371 @defmac STACK_REG_COVER_CLASS
2372 This is a cover class containing the stack registers. Define this if
2373 the machine has any stack-like registers.
2376 @defmac FIRST_STACK_REG
2377 The number of the first stack-like register. This one is the top
2381 @defmac LAST_STACK_REG
2382 The number of the last stack-like register. This one is the bottom of
2386 @node Register Classes
2387 @section Register Classes
2388 @cindex register class definitions
2389 @cindex class definitions, register
2391 On many machines, the numbered registers are not all equivalent.
2392 For example, certain registers may not be allowed for indexed addressing;
2393 certain registers may not be allowed in some instructions. These machine
2394 restrictions are described to the compiler using @dfn{register classes}.
2396 You define a number of register classes, giving each one a name and saying
2397 which of the registers belong to it. Then you can specify register classes
2398 that are allowed as operands to particular instruction patterns.
2402 In general, each register will belong to several classes. In fact, one
2403 class must be named @code{ALL_REGS} and contain all the registers. Another
2404 class must be named @code{NO_REGS} and contain no registers. Often the
2405 union of two classes will be another class; however, this is not required.
2407 @findex GENERAL_REGS
2408 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2409 terribly special about the name, but the operand constraint letters
2410 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2411 the same as @code{ALL_REGS}, just define it as a macro which expands
2414 Order the classes so that if class @var{x} is contained in class @var{y}
2415 then @var{x} has a lower class number than @var{y}.
2417 The way classes other than @code{GENERAL_REGS} are specified in operand
2418 constraints is through machine-dependent operand constraint letters.
2419 You can define such letters to correspond to various classes, then use
2420 them in operand constraints.
2422 You should define a class for the union of two classes whenever some
2423 instruction allows both classes. For example, if an instruction allows
2424 either a floating point (coprocessor) register or a general register for a
2425 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2426 which includes both of them. Otherwise you will get suboptimal code.
2428 You must also specify certain redundant information about the register
2429 classes: for each class, which classes contain it and which ones are
2430 contained in it; for each pair of classes, the largest class contained
2433 When a value occupying several consecutive registers is expected in a
2434 certain class, all the registers used must belong to that class.
2435 Therefore, register classes cannot be used to enforce a requirement for
2436 a register pair to start with an even-numbered register. The way to
2437 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2439 Register classes used for input-operands of bitwise-and or shift
2440 instructions have a special requirement: each such class must have, for
2441 each fixed-point machine mode, a subclass whose registers can transfer that
2442 mode to or from memory. For example, on some machines, the operations for
2443 single-byte values (@code{QImode}) are limited to certain registers. When
2444 this is so, each register class that is used in a bitwise-and or shift
2445 instruction must have a subclass consisting of registers from which
2446 single-byte values can be loaded or stored. This is so that
2447 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2449 @deftp {Data type} {enum reg_class}
2450 An enumerated type that must be defined with all the register class names
2451 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2452 must be the last register class, followed by one more enumerated value,
2453 @code{LIM_REG_CLASSES}, which is not a register class but rather
2454 tells how many classes there are.
2456 Each register class has a number, which is the value of casting
2457 the class name to type @code{int}. The number serves as an index
2458 in many of the tables described below.
2461 @defmac N_REG_CLASSES
2462 The number of distinct register classes, defined as follows:
2465 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2469 @defmac REG_CLASS_NAMES
2470 An initializer containing the names of the register classes as C string
2471 constants. These names are used in writing some of the debugging dumps.
2474 @defmac REG_CLASS_CONTENTS
2475 An initializer containing the contents of the register classes, as integers
2476 which are bit masks. The @var{n}th integer specifies the contents of class
2477 @var{n}. The way the integer @var{mask} is interpreted is that
2478 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2480 When the machine has more than 32 registers, an integer does not suffice.
2481 Then the integers are replaced by sub-initializers, braced groupings containing
2482 several integers. Each sub-initializer must be suitable as an initializer
2483 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2484 In this situation, the first integer in each sub-initializer corresponds to
2485 registers 0 through 31, the second integer to registers 32 through 63, and
2489 @defmac REGNO_REG_CLASS (@var{regno})
2490 A C expression whose value is a register class containing hard register
2491 @var{regno}. In general there is more than one such class; choose a class
2492 which is @dfn{minimal}, meaning that no smaller class also contains the
2496 @defmac BASE_REG_CLASS
2497 A macro whose definition is the name of the class to which a valid
2498 base register must belong. A base register is one used in an address
2499 which is the register value plus a displacement.
2502 @defmac MODE_BASE_REG_CLASS (@var{mode})
2503 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2504 the selection of a base register in a mode dependent manner. If
2505 @var{mode} is VOIDmode then it should return the same value as
2506 @code{BASE_REG_CLASS}.
2509 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2510 A C expression whose value is the register class to which a valid
2511 base register must belong in order to be used in a base plus index
2512 register address. You should define this macro if base plus index
2513 addresses have different requirements than other base register uses.
2516 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{outer_code}, @var{index_code})
2517 A C expression whose value is the register class to which a valid
2518 base register must belong. @var{outer_code} and @var{index_code} define the
2519 context in which the base register occurs. @var{outer_code} is the code of
2520 the immediately enclosing expression (@code{MEM} for the top level of an
2521 address, @code{ADDRESS} for something that occurs in an
2522 @code{address_operand}). @var{index_code} is the code of the corresponding
2523 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2526 @defmac INDEX_REG_CLASS
2527 A macro whose definition is the name of the class to which a valid
2528 index register must belong. An index register is one used in an
2529 address where its value is either multiplied by a scale factor or
2530 added to another register (as well as added to a displacement).
2533 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2534 A C expression which is nonzero if register number @var{num} is
2535 suitable for use as a base register in operand addresses.
2536 Like @code{TARGET_LEGITIMATE_ADDRESS_P}, this macro should also
2537 define a strict and a non-strict variant. Both variants behave
2538 the same for hard register; for pseudos, the strict variant will
2539 pass only those that have been allocated to a valid hard registers,
2540 while the non-strict variant will pass all pseudos.
2542 @findex REG_OK_STRICT
2543 Compiler source files that want to use the strict variant of this and
2544 other macros define the macro @code{REG_OK_STRICT}. You should use an
2545 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
2546 that case and the non-strict variant otherwise.
2549 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2550 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2551 that expression may examine the mode of the memory reference in
2552 @var{mode}. You should define this macro if the mode of the memory
2553 reference affects whether a register may be used as a base register. If
2554 you define this macro, the compiler will use it instead of
2555 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2556 addresses that appear outside a @code{MEM}, i.e., as an
2557 @code{address_operand}.
2559 This macro also has strict and non-strict variants.
2562 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2563 A C expression which is nonzero if register number @var{num} is suitable for
2564 use as a base register in base plus index operand addresses, accessing
2565 memory in mode @var{mode}. It may be either a suitable hard register or a
2566 pseudo register that has been allocated such a hard register. You should
2567 define this macro if base plus index addresses have different requirements
2568 than other base register uses.
2570 Use of this macro is deprecated; please use the more general
2571 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2573 This macro also has strict and non-strict variants.
2576 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{outer_code}, @var{index_code})
2577 A C expression that is just like @code{REGNO_MODE_OK_FOR_BASE_P}, except
2578 that that expression may examine the context in which the register
2579 appears in the memory reference. @var{outer_code} is the code of the
2580 immediately enclosing expression (@code{MEM} if at the top level of the
2581 address, @code{ADDRESS} for something that occurs in an
2582 @code{address_operand}). @var{index_code} is the code of the
2583 corresponding index expression if @var{outer_code} is @code{PLUS};
2584 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2585 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2587 This macro also has strict and non-strict variants.
2590 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2591 A C expression which is nonzero if register number @var{num} is
2592 suitable for use as an index register in operand addresses. It may be
2593 either a suitable hard register or a pseudo register that has been
2594 allocated such a hard register.
2596 The difference between an index register and a base register is that
2597 the index register may be scaled. If an address involves the sum of
2598 two registers, neither one of them scaled, then either one may be
2599 labeled the ``base'' and the other the ``index''; but whichever
2600 labeling is used must fit the machine's constraints of which registers
2601 may serve in each capacity. The compiler will try both labelings,
2602 looking for one that is valid, and will reload one or both registers
2603 only if neither labeling works.
2605 This macro also has strict and non-strict variants.
2608 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2609 A C expression that places additional restrictions on the register class
2610 to use when it is necessary to copy value @var{x} into a register in class
2611 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2612 another, smaller class. On many machines, the following definition is
2616 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2619 Sometimes returning a more restrictive class makes better code. For
2620 example, on the 68000, when @var{x} is an integer constant that is in range
2621 for a @samp{moveq} instruction, the value of this macro is always
2622 @code{DATA_REGS} as long as @var{class} includes the data registers.
2623 Requiring a data register guarantees that a @samp{moveq} will be used.
2625 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2626 @var{class} is if @var{x} is a legitimate constant which cannot be
2627 loaded into some register class. By returning @code{NO_REGS} you can
2628 force @var{x} into a memory location. For example, rs6000 can load
2629 immediate values into general-purpose registers, but does not have an
2630 instruction for loading an immediate value into a floating-point
2631 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2632 @var{x} is a floating-point constant. If the constant can't be loaded
2633 into any kind of register, code generation will be better if
2634 @code{LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2635 of using @code{PREFERRED_RELOAD_CLASS}.
2637 If an insn has pseudos in it after register allocation, reload will go
2638 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2639 to find the best one. Returning @code{NO_REGS}, in this case, makes
2640 reload add a @code{!} in front of the constraint: the x86 back-end uses
2641 this feature to discourage usage of 387 registers when math is done in
2642 the SSE registers (and vice versa).
2645 @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2646 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2647 input reloads. If you don't define this macro, the default is to use
2648 @var{class}, unchanged.
2650 You can also use @code{PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2651 reload from using some alternatives, like @code{PREFERRED_RELOAD_CLASS}.
2654 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2655 A C expression that places additional restrictions on the register class
2656 to use when it is necessary to be able to hold a value of mode
2657 @var{mode} in a reload register for which class @var{class} would
2660 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2661 there are certain modes that simply can't go in certain reload classes.
2663 The value is a register class; perhaps @var{class}, or perhaps another,
2666 Don't define this macro unless the target machine has limitations which
2667 require the macro to do something nontrivial.
2670 @hook TARGET_SECONDARY_RELOAD
2671 Many machines have some registers that cannot be copied directly to or
2672 from memory or even from other types of registers. An example is the
2673 @samp{MQ} register, which on most machines, can only be copied to or
2674 from general registers, but not memory. Below, we shall be using the
2675 term 'intermediate register' when a move operation cannot be performed
2676 directly, but has to be done by copying the source into the intermediate
2677 register first, and then copying the intermediate register to the
2678 destination. An intermediate register always has the same mode as
2679 source and destination. Since it holds the actual value being copied,
2680 reload might apply optimizations to re-use an intermediate register
2681 and eliding the copy from the source when it can determine that the
2682 intermediate register still holds the required value.
2684 Another kind of secondary reload is required on some machines which
2685 allow copying all registers to and from memory, but require a scratch
2686 register for stores to some memory locations (e.g., those with symbolic
2687 address on the RT, and those with certain symbolic address on the SPARC
2688 when compiling PIC)@. Scratch registers need not have the same mode
2689 as the value being copied, and usually hold a different value than
2690 that being copied. Special patterns in the md file are needed to
2691 describe how the copy is performed with the help of the scratch register;
2692 these patterns also describe the number, register class(es) and mode(s)
2693 of the scratch register(s).
2695 In some cases, both an intermediate and a scratch register are required.
2697 For input reloads, this target hook is called with nonzero @var{in_p},
2698 and @var{x} is an rtx that needs to be copied to a register of class
2699 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2700 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2701 needs to be copied to rtx @var{x} in @var{reload_mode}.
2703 If copying a register of @var{reload_class} from/to @var{x} requires
2704 an intermediate register, the hook @code{secondary_reload} should
2705 return the register class required for this intermediate register.
2706 If no intermediate register is required, it should return NO_REGS.
2707 If more than one intermediate register is required, describe the one
2708 that is closest in the copy chain to the reload register.
2710 If scratch registers are needed, you also have to describe how to
2711 perform the copy from/to the reload register to/from this
2712 closest intermediate register. Or if no intermediate register is
2713 required, but still a scratch register is needed, describe the
2714 copy from/to the reload register to/from the reload operand @var{x}.
2716 You do this by setting @code{sri->icode} to the instruction code of a pattern
2717 in the md file which performs the move. Operands 0 and 1 are the output
2718 and input of this copy, respectively. Operands from operand 2 onward are
2719 for scratch operands. These scratch operands must have a mode, and a
2720 single-register-class
2721 @c [later: or memory]
2724 When an intermediate register is used, the @code{secondary_reload}
2725 hook will be called again to determine how to copy the intermediate
2726 register to/from the reload operand @var{x}, so your hook must also
2727 have code to handle the register class of the intermediate operand.
2729 @c [For later: maybe we'll allow multi-alternative reload patterns -
2730 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2731 @c and match the constraints of input and output to determine the required
2732 @c alternative. A restriction would be that constraints used to match
2733 @c against reloads registers would have to be written as register class
2734 @c constraints, or we need a new target macro / hook that tells us if an
2735 @c arbitrary constraint can match an unknown register of a given class.
2736 @c Such a macro / hook would also be useful in other places.]
2739 @var{x} might be a pseudo-register or a @code{subreg} of a
2740 pseudo-register, which could either be in a hard register or in memory.
2741 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2742 in memory and the hard register number if it is in a register.
2744 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2745 currently not supported. For the time being, you will have to continue
2746 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2748 @code{copy_cost} also uses this target hook to find out how values are
2749 copied. If you want it to include some extra cost for the need to allocate
2750 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2751 Or if two dependent moves are supposed to have a lower cost than the sum
2752 of the individual moves due to expected fortuitous scheduling and/or special
2753 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2756 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2757 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2758 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2759 These macros are obsolete, new ports should use the target hook
2760 @code{TARGET_SECONDARY_RELOAD} instead.
2762 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2763 target hook. Older ports still define these macros to indicate to the
2764 reload phase that it may
2765 need to allocate at least one register for a reload in addition to the
2766 register to contain the data. Specifically, if copying @var{x} to a
2767 register @var{class} in @var{mode} requires an intermediate register,
2768 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2769 largest register class all of whose registers can be used as
2770 intermediate registers or scratch registers.
2772 If copying a register @var{class} in @var{mode} to @var{x} requires an
2773 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2774 was supposed to be defined be defined to return the largest register
2775 class required. If the
2776 requirements for input and output reloads were the same, the macro
2777 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2780 The values returned by these macros are often @code{GENERAL_REGS}.
2781 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2782 can be directly copied to or from a register of @var{class} in
2783 @var{mode} without requiring a scratch register. Do not define this
2784 macro if it would always return @code{NO_REGS}.
2786 If a scratch register is required (either with or without an
2787 intermediate register), you were supposed to define patterns for
2788 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2789 (@pxref{Standard Names}. These patterns, which were normally
2790 implemented with a @code{define_expand}, should be similar to the
2791 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2794 These patterns need constraints for the reload register and scratch
2796 contain a single register class. If the original reload register (whose
2797 class is @var{class}) can meet the constraint given in the pattern, the
2798 value returned by these macros is used for the class of the scratch
2799 register. Otherwise, two additional reload registers are required.
2800 Their classes are obtained from the constraints in the insn pattern.
2802 @var{x} might be a pseudo-register or a @code{subreg} of a
2803 pseudo-register, which could either be in a hard register or in memory.
2804 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2805 in memory and the hard register number if it is in a register.
2807 These macros should not be used in the case where a particular class of
2808 registers can only be copied to memory and not to another class of
2809 registers. In that case, secondary reload registers are not needed and
2810 would not be helpful. Instead, a stack location must be used to perform
2811 the copy and the @code{mov@var{m}} pattern should use memory as an
2812 intermediate storage. This case often occurs between floating-point and
2816 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2817 Certain machines have the property that some registers cannot be copied
2818 to some other registers without using memory. Define this macro on
2819 those machines to be a C expression that is nonzero if objects of mode
2820 @var{m} in registers of @var{class1} can only be copied to registers of
2821 class @var{class2} by storing a register of @var{class1} into memory
2822 and loading that memory location into a register of @var{class2}.
2824 Do not define this macro if its value would always be zero.
2827 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2828 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2829 allocates a stack slot for a memory location needed for register copies.
2830 If this macro is defined, the compiler instead uses the memory location
2831 defined by this macro.
2833 Do not define this macro if you do not define
2834 @code{SECONDARY_MEMORY_NEEDED}.
2837 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2838 When the compiler needs a secondary memory location to copy between two
2839 registers of mode @var{mode}, it normally allocates sufficient memory to
2840 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2841 load operations in a mode that many bits wide and whose class is the
2842 same as that of @var{mode}.
2844 This is right thing to do on most machines because it ensures that all
2845 bits of the register are copied and prevents accesses to the registers
2846 in a narrower mode, which some machines prohibit for floating-point
2849 However, this default behavior is not correct on some machines, such as
2850 the DEC Alpha, that store short integers in floating-point registers
2851 differently than in integer registers. On those machines, the default
2852 widening will not work correctly and you must define this macro to
2853 suppress that widening in some cases. See the file @file{alpha.h} for
2856 Do not define this macro if you do not define
2857 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2858 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2861 @defmac CLASS_LIKELY_SPILLED_P (@var{class})
2862 A C expression whose value is nonzero if pseudos that have been assigned
2863 to registers of class @var{class} would likely be spilled because
2864 registers of @var{class} are needed for spill registers.
2866 The default value of this macro returns 1 if @var{class} has exactly one
2867 register and zero otherwise. On most machines, this default should be
2868 used. Only define this macro to some other expression if pseudos
2869 allocated by @file{local-alloc.c} end up in memory because their hard
2870 registers were needed for spill registers. If this macro returns nonzero
2871 for those classes, those pseudos will only be allocated by
2872 @file{global.c}, which knows how to reallocate the pseudo to another
2873 register. If there would not be another register available for
2874 reallocation, you should not change the definition of this macro since
2875 the only effect of such a definition would be to slow down register
2879 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2880 A C expression for the maximum number of consecutive registers
2881 of class @var{class} needed to hold a value of mode @var{mode}.
2883 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2884 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2885 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2886 @var{mode})} for all @var{regno} values in the class @var{class}.
2888 This macro helps control the handling of multiple-word values
2892 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2893 If defined, a C expression that returns nonzero for a @var{class} for which
2894 a change from mode @var{from} to mode @var{to} is invalid.
2896 For the example, loading 32-bit integer or floating-point objects into
2897 floating-point registers on the Alpha extends them to 64 bits.
2898 Therefore loading a 64-bit object and then storing it as a 32-bit object
2899 does not store the low-order 32 bits, as would be the case for a normal
2900 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2904 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2905 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2906 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2910 @hook TARGET_IRA_COVER_CLASSES
2911 Return an array of cover classes for the Integrated Register Allocator
2912 (@acronym{IRA}). Cover classes are a set of non-intersecting register
2913 classes covering all hard registers used for register allocation
2914 purposes. If a move between two registers in the same cover class is
2915 possible, it should be cheaper than a load or store of the registers.
2916 The array is terminated by a @code{LIM_REG_CLASSES} element.
2918 The order of cover classes in the array is important. If two classes
2919 have the same cost of usage for a pseudo, the class occurred first in
2920 the array is chosen for the pseudo.
2922 This hook is called once at compiler startup, after the command-line
2923 options have been processed. It is then re-examined by every call to
2924 @code{target_reinit}.
2926 The default implementation returns @code{IRA_COVER_CLASSES}, if defined,
2927 otherwise there is no default implementation. You must define either this
2928 macro or @code{IRA_COVER_CLASSES} in order to use the integrated register
2929 allocator with Chaitin-Briggs coloring. If the macro is not defined,
2930 the only available coloring algorithm is Chow's priority coloring.
2933 @defmac IRA_COVER_CLASSES
2934 See the documentation for @code{TARGET_IRA_COVER_CLASSES}.
2937 @node Old Constraints
2938 @section Obsolete Macros for Defining Constraints
2939 @cindex defining constraints, obsolete method
2940 @cindex constraints, defining, obsolete method
2942 Machine-specific constraints can be defined with these macros instead
2943 of the machine description constructs described in @ref{Define
2944 Constraints}. This mechanism is obsolete. New ports should not use
2945 it; old ports should convert to the new mechanism.
2947 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2948 For the constraint at the start of @var{str}, which starts with the letter
2949 @var{c}, return the length. This allows you to have register class /
2950 constant / extra constraints that are longer than a single letter;
2951 you don't need to define this macro if you can do with single-letter
2952 constraints only. The definition of this macro should use
2953 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2954 to handle specially.
2955 There are some sanity checks in genoutput.c that check the constraint lengths
2956 for the md file, so you can also use this macro to help you while you are
2957 transitioning from a byzantine single-letter-constraint scheme: when you
2958 return a negative length for a constraint you want to re-use, genoutput
2959 will complain about every instance where it is used in the md file.
2962 @defmac REG_CLASS_FROM_LETTER (@var{char})
2963 A C expression which defines the machine-dependent operand constraint
2964 letters for register classes. If @var{char} is such a letter, the
2965 value should be the register class corresponding to it. Otherwise,
2966 the value should be @code{NO_REGS}. The register letter @samp{r},
2967 corresponding to class @code{GENERAL_REGS}, will not be passed
2968 to this macro; you do not need to handle it.
2971 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2972 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2973 passed in @var{str}, so that you can use suffixes to distinguish between
2977 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2978 A C expression that defines the machine-dependent operand constraint
2979 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2980 particular ranges of integer values. If @var{c} is one of those
2981 letters, the expression should check that @var{value}, an integer, is in
2982 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2983 not one of those letters, the value should be 0 regardless of
2987 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2988 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2989 string passed in @var{str}, so that you can use suffixes to distinguish
2990 between different variants.
2993 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2994 A C expression that defines the machine-dependent operand constraint
2995 letters that specify particular ranges of @code{const_double} values
2996 (@samp{G} or @samp{H}).
2998 If @var{c} is one of those letters, the expression should check that
2999 @var{value}, an RTX of code @code{const_double}, is in the appropriate
3000 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
3001 letters, the value should be 0 regardless of @var{value}.
3003 @code{const_double} is used for all floating-point constants and for
3004 @code{DImode} fixed-point constants. A given letter can accept either
3005 or both kinds of values. It can use @code{GET_MODE} to distinguish
3006 between these kinds.
3009 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
3010 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
3011 string passed in @var{str}, so that you can use suffixes to distinguish
3012 between different variants.
3015 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
3016 A C expression that defines the optional machine-dependent constraint
3017 letters that can be used to segregate specific types of operands, usually
3018 memory references, for the target machine. Any letter that is not
3019 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
3020 @code{REG_CLASS_FROM_CONSTRAINT}
3021 may be used. Normally this macro will not be defined.
3023 If it is required for a particular target machine, it should return 1
3024 if @var{value} corresponds to the operand type represented by the
3025 constraint letter @var{c}. If @var{c} is not defined as an extra
3026 constraint, the value returned should be 0 regardless of @var{value}.
3028 For example, on the ROMP, load instructions cannot have their output
3029 in r0 if the memory reference contains a symbolic address. Constraint
3030 letter @samp{Q} is defined as representing a memory address that does
3031 @emph{not} contain a symbolic address. An alternative is specified with
3032 a @samp{Q} constraint on the input and @samp{r} on the output. The next
3033 alternative specifies @samp{m} on the input and a register class that
3034 does not include r0 on the output.
3037 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
3038 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
3039 in @var{str}, so that you can use suffixes to distinguish between different
3043 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
3044 A C expression that defines the optional machine-dependent constraint
3045 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
3046 be treated like memory constraints by the reload pass.
3048 It should return 1 if the operand type represented by the constraint
3049 at the start of @var{str}, the first letter of which is the letter @var{c},
3050 comprises a subset of all memory references including
3051 all those whose address is simply a base register. This allows the reload
3052 pass to reload an operand, if it does not directly correspond to the operand
3053 type of @var{c}, by copying its address into a base register.
3055 For example, on the S/390, some instructions do not accept arbitrary
3056 memory references, but only those that do not make use of an index
3057 register. The constraint letter @samp{Q} is defined via
3058 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
3059 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
3060 a @samp{Q} constraint can handle any memory operand, because the
3061 reload pass knows it can be reloaded by copying the memory address
3062 into a base register if required. This is analogous to the way
3063 an @samp{o} constraint can handle any memory operand.
3066 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
3067 A C expression that defines the optional machine-dependent constraint
3068 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
3069 @code{EXTRA_CONSTRAINT_STR}, that should
3070 be treated like address constraints by the reload pass.
3072 It should return 1 if the operand type represented by the constraint
3073 at the start of @var{str}, which starts with the letter @var{c}, comprises
3074 a subset of all memory addresses including
3075 all those that consist of just a base register. This allows the reload
3076 pass to reload an operand, if it does not directly correspond to the operand
3077 type of @var{str}, by copying it into a base register.
3079 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3080 be used with the @code{address_operand} predicate. It is treated
3081 analogously to the @samp{p} constraint.
3084 @node Stack and Calling
3085 @section Stack Layout and Calling Conventions
3086 @cindex calling conventions
3088 @c prevent bad page break with this line
3089 This describes the stack layout and calling conventions.
3093 * Exception Handling::
3098 * Register Arguments::
3100 * Aggregate Return::
3105 * Stack Smashing Protection::
3109 @subsection Basic Stack Layout
3110 @cindex stack frame layout
3111 @cindex frame layout
3113 @c prevent bad page break with this line
3114 Here is the basic stack layout.
3116 @defmac STACK_GROWS_DOWNWARD
3117 Define this macro if pushing a word onto the stack moves the stack
3118 pointer to a smaller address.
3120 When we say, ``define this macro if @dots{}'', it means that the
3121 compiler checks this macro only with @code{#ifdef} so the precise
3122 definition used does not matter.
3125 @defmac STACK_PUSH_CODE
3126 This macro defines the operation used when something is pushed
3127 on the stack. In RTL, a push operation will be
3128 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3130 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3131 and @code{POST_INC}. Which of these is correct depends on
3132 the stack direction and on whether the stack pointer points
3133 to the last item on the stack or whether it points to the
3134 space for the next item on the stack.
3136 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3137 defined, which is almost always right, and @code{PRE_INC} otherwise,
3138 which is often wrong.
3141 @defmac FRAME_GROWS_DOWNWARD
3142 Define this macro to nonzero value if the addresses of local variable slots
3143 are at negative offsets from the frame pointer.
3146 @defmac ARGS_GROW_DOWNWARD
3147 Define this macro if successive arguments to a function occupy decreasing
3148 addresses on the stack.
3151 @defmac STARTING_FRAME_OFFSET
3152 Offset from the frame pointer to the first local variable slot to be allocated.
3154 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3155 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3156 Otherwise, it is found by adding the length of the first slot to the
3157 value @code{STARTING_FRAME_OFFSET}.
3158 @c i'm not sure if the above is still correct.. had to change it to get
3159 @c rid of an overfull. --mew 2feb93
3162 @defmac STACK_ALIGNMENT_NEEDED
3163 Define to zero to disable final alignment of the stack during reload.
3164 The nonzero default for this macro is suitable for most ports.
3166 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3167 is a register save block following the local block that doesn't require
3168 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3169 stack alignment and do it in the backend.
3172 @defmac STACK_POINTER_OFFSET
3173 Offset from the stack pointer register to the first location at which
3174 outgoing arguments are placed. If not specified, the default value of
3175 zero is used. This is the proper value for most machines.
3177 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3178 the first location at which outgoing arguments are placed.
3181 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3182 Offset from the argument pointer register to the first argument's
3183 address. On some machines it may depend on the data type of the
3186 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3187 the first argument's address.
3190 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3191 Offset from the stack pointer register to an item dynamically allocated
3192 on the stack, e.g., by @code{alloca}.
3194 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3195 length of the outgoing arguments. The default is correct for most
3196 machines. See @file{function.c} for details.
3199 @defmac INITIAL_FRAME_ADDRESS_RTX
3200 A C expression whose value is RTL representing the address of the initial
3201 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3202 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3203 default value will be used. Define this macro in order to make frame pointer
3204 elimination work in the presence of @code{__builtin_frame_address (count)} and
3205 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3208 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3209 A C expression whose value is RTL representing the address in a stack
3210 frame where the pointer to the caller's frame is stored. Assume that
3211 @var{frameaddr} is an RTL expression for the address of the stack frame
3214 If you don't define this macro, the default is to return the value
3215 of @var{frameaddr}---that is, the stack frame address is also the
3216 address of the stack word that points to the previous frame.
3219 @defmac SETUP_FRAME_ADDRESSES
3220 If defined, a C expression that produces the machine-specific code to
3221 setup the stack so that arbitrary frames can be accessed. For example,
3222 on the SPARC, we must flush all of the register windows to the stack
3223 before we can access arbitrary stack frames. You will seldom need to
3227 @hook TARGET_BUILTIN_SETJMP_FRAME_VALUE
3228 This target hook should return an rtx that is used to store
3229 the address of the current frame into the built in @code{setjmp} buffer.
3230 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3231 machines. One reason you may need to define this target hook is if
3232 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3235 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3236 A C expression whose value is RTL representing the value of the frame
3237 address for the current frame. @var{frameaddr} is the frame pointer
3238 of the current frame. This is used for __builtin_frame_address.
3239 You need only define this macro if the frame address is not the same
3240 as the frame pointer. Most machines do not need to define it.
3243 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3244 A C expression whose value is RTL representing the value of the return
3245 address for the frame @var{count} steps up from the current frame, after
3246 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3247 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3248 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3250 The value of the expression must always be the correct address when
3251 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3252 determine the return address of other frames.
3255 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3256 Define this if the return address of a particular stack frame is accessed
3257 from the frame pointer of the previous stack frame.
3260 @defmac INCOMING_RETURN_ADDR_RTX
3261 A C expression whose value is RTL representing the location of the
3262 incoming return address at the beginning of any function, before the
3263 prologue. This RTL is either a @code{REG}, indicating that the return
3264 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3267 You only need to define this macro if you want to support call frame
3268 debugging information like that provided by DWARF 2.
3270 If this RTL is a @code{REG}, you should also define
3271 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3274 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3275 A C expression whose value is an integer giving a DWARF 2 column
3276 number that may be used as an alternative return column. The column
3277 must not correspond to any gcc hard register (that is, it must not
3278 be in the range of @code{DWARF_FRAME_REGNUM}).
3280 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3281 general register, but an alternative column needs to be used for signal
3282 frames. Some targets have also used different frame return columns
3286 @defmac DWARF_ZERO_REG
3287 A C expression whose value is an integer giving a DWARF 2 register
3288 number that is considered to always have the value zero. This should
3289 only be defined if the target has an architected zero register, and
3290 someone decided it was a good idea to use that register number to
3291 terminate the stack backtrace. New ports should avoid this.
3294 @hook TARGET_DWARF_HANDLE_FRAME_UNSPEC
3295 This target hook allows the backend to emit frame-related insns that
3296 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3297 info engine will invoke it on insns of the form
3299 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3303 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3305 to let the backend emit the call frame instructions. @var{label} is
3306 the CFI label attached to the insn, @var{pattern} is the pattern of
3307 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3310 @defmac INCOMING_FRAME_SP_OFFSET
3311 A C expression whose value is an integer giving the offset, in bytes,
3312 from the value of the stack pointer register to the top of the stack
3313 frame at the beginning of any function, before the prologue. The top of
3314 the frame is defined to be the value of the stack pointer in the
3315 previous frame, just before the call instruction.
3317 You only need to define this macro if you want to support call frame
3318 debugging information like that provided by DWARF 2.
3321 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3322 A C expression whose value is an integer giving the offset, in bytes,
3323 from the argument pointer to the canonical frame address (cfa). The
3324 final value should coincide with that calculated by
3325 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3326 during virtual register instantiation.
3328 The default value for this macro is
3329 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
3330 which is correct for most machines; in general, the arguments are found
3331 immediately before the stack frame. Note that this is not the case on
3332 some targets that save registers into the caller's frame, such as SPARC
3333 and rs6000, and so such targets need to define this macro.
3335 You only need to define this macro if the default is incorrect, and you
3336 want to support call frame debugging information like that provided by
3340 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3341 If defined, a C expression whose value is an integer giving the offset
3342 in bytes from the frame pointer to the canonical frame address (cfa).
3343 The final value should coincide with that calculated by
3344 @code{INCOMING_FRAME_SP_OFFSET}.
3346 Normally the CFA is calculated as an offset from the argument pointer,
3347 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3348 variable due to the ABI, this may not be possible. If this macro is
3349 defined, it implies that the virtual register instantiation should be
3350 based on the frame pointer instead of the argument pointer. Only one
3351 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3355 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3356 If defined, a C expression whose value is an integer giving the offset
3357 in bytes from the canonical frame address (cfa) to the frame base used
3358 in DWARF 2 debug information. The default is zero. A different value
3359 may reduce the size of debug information on some ports.
3362 @node Exception Handling
3363 @subsection Exception Handling Support
3364 @cindex exception handling
3366 @defmac EH_RETURN_DATA_REGNO (@var{N})
3367 A C expression whose value is the @var{N}th register number used for
3368 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3369 @var{N} registers are usable.
3371 The exception handling library routines communicate with the exception
3372 handlers via a set of agreed upon registers. Ideally these registers
3373 should be call-clobbered; it is possible to use call-saved registers,
3374 but may negatively impact code size. The target must support at least
3375 2 data registers, but should define 4 if there are enough free registers.
3377 You must define this macro if you want to support call frame exception
3378 handling like that provided by DWARF 2.
3381 @defmac EH_RETURN_STACKADJ_RTX
3382 A C expression whose value is RTL representing a location in which
3383 to store a stack adjustment to be applied before function return.
3384 This is used to unwind the stack to an exception handler's call frame.
3385 It will be assigned zero on code paths that return normally.
3387 Typically this is a call-clobbered hard register that is otherwise
3388 untouched by the epilogue, but could also be a stack slot.
3390 Do not define this macro if the stack pointer is saved and restored
3391 by the regular prolog and epilog code in the call frame itself; in
3392 this case, the exception handling library routines will update the
3393 stack location to be restored in place. Otherwise, you must define
3394 this macro if you want to support call frame exception handling like
3395 that provided by DWARF 2.
3398 @defmac EH_RETURN_HANDLER_RTX
3399 A C expression whose value is RTL representing a location in which
3400 to store the address of an exception handler to which we should
3401 return. It will not be assigned on code paths that return normally.
3403 Typically this is the location in the call frame at which the normal
3404 return address is stored. For targets that return by popping an
3405 address off the stack, this might be a memory address just below
3406 the @emph{target} call frame rather than inside the current call
3407 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3408 been assigned, so it may be used to calculate the location of the
3411 Some targets have more complex requirements than storing to an
3412 address calculable during initial code generation. In that case
3413 the @code{eh_return} instruction pattern should be used instead.
3415 If you want to support call frame exception handling, you must
3416 define either this macro or the @code{eh_return} instruction pattern.
3419 @defmac RETURN_ADDR_OFFSET
3420 If defined, an integer-valued C expression for which rtl will be generated
3421 to add it to the exception handler address before it is searched in the
3422 exception handling tables, and to subtract it again from the address before
3423 using it to return to the exception handler.
3426 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3427 This macro chooses the encoding of pointers embedded in the exception
3428 handling sections. If at all possible, this should be defined such
3429 that the exception handling section will not require dynamic relocations,
3430 and so may be read-only.
3432 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3433 @var{global} is true if the symbol may be affected by dynamic relocations.
3434 The macro should return a combination of the @code{DW_EH_PE_*} defines
3435 as found in @file{dwarf2.h}.
3437 If this macro is not defined, pointers will not be encoded but
3438 represented directly.
3441 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3442 This macro allows the target to emit whatever special magic is required
3443 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3444 Generic code takes care of pc-relative and indirect encodings; this must
3445 be defined if the target uses text-relative or data-relative encodings.
3447 This is a C statement that branches to @var{done} if the format was
3448 handled. @var{encoding} is the format chosen, @var{size} is the number
3449 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3453 @defmac MD_UNWIND_SUPPORT
3454 A string specifying a file to be #include'd in unwind-dw2.c. The file
3455 so included typically defines @code{MD_FALLBACK_FRAME_STATE_FOR}.
3458 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3459 This macro allows the target to add CPU and operating system specific
3460 code to the call-frame unwinder for use when there is no unwind data
3461 available. The most common reason to implement this macro is to unwind
3462 through signal frames.
3464 This macro is called from @code{uw_frame_state_for} in
3465 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3466 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3467 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3468 for the address of the code being executed and @code{context->cfa} for
3469 the stack pointer value. If the frame can be decoded, the register
3470 save addresses should be updated in @var{fs} and the macro should
3471 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3472 the macro should evaluate to @code{_URC_END_OF_STACK}.
3474 For proper signal handling in Java this macro is accompanied by
3475 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3478 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3479 This macro allows the target to add operating system specific code to the
3480 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3481 usually used for signal or interrupt frames.
3483 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3484 @var{context} is an @code{_Unwind_Context};
3485 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3486 for the abi and context in the @code{.unwabi} directive. If the
3487 @code{.unwabi} directive can be handled, the register save addresses should
3488 be updated in @var{fs}.
3491 @defmac TARGET_USES_WEAK_UNWIND_INFO
3492 A C expression that evaluates to true if the target requires unwind
3493 info to be given comdat linkage. Define it to be @code{1} if comdat
3494 linkage is necessary. The default is @code{0}.
3497 @node Stack Checking
3498 @subsection Specifying How Stack Checking is Done
3500 GCC will check that stack references are within the boundaries of the
3501 stack, if the option @option{-fstack-check} is specified, in one of
3506 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3507 will assume that you have arranged for full stack checking to be done
3508 at appropriate places in the configuration files. GCC will not do
3509 other special processing.
3512 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3513 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3514 that you have arranged for static stack checking (checking of the
3515 static stack frame of functions) to be done at appropriate places
3516 in the configuration files. GCC will only emit code to do dynamic
3517 stack checking (checking on dynamic stack allocations) using the third
3521 If neither of the above are true, GCC will generate code to periodically
3522 ``probe'' the stack pointer using the values of the macros defined below.
3525 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3526 GCC will change its allocation strategy for large objects if the option
3527 @option{-fstack-check} is specified: they will always be allocated
3528 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3530 @defmac STACK_CHECK_BUILTIN
3531 A nonzero value if stack checking is done by the configuration files in a
3532 machine-dependent manner. You should define this macro if stack checking
3533 is required by the ABI of your machine or if you would like to do stack
3534 checking in some more efficient way than the generic approach. The default
3535 value of this macro is zero.
3538 @defmac STACK_CHECK_STATIC_BUILTIN
3539 A nonzero value if static stack checking is done by the configuration files
3540 in a machine-dependent manner. You should define this macro if you would
3541 like to do static stack checking in some more efficient way than the generic
3542 approach. The default value of this macro is zero.
3545 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
3546 An integer specifying the interval at which GCC must generate stack probe
3547 instructions, defined as 2 raised to this integer. You will normally
3548 define this macro so that the interval be no larger than the size of
3549 the ``guard pages'' at the end of a stack area. The default value
3550 of 12 (4096-byte interval) is suitable for most systems.
3553 @defmac STACK_CHECK_MOVING_SP
3554 An integer which is nonzero if GCC should move the stack pointer page by page
3555 when doing probes. This can be necessary on systems where the stack pointer
3556 contains the bottom address of the memory area accessible to the executing
3557 thread at any point in time. In this situation an alternate signal stack
3558 is required in order to be able to recover from a stack overflow. The
3559 default value of this macro is zero.
3562 @defmac STACK_CHECK_PROTECT
3563 The number of bytes of stack needed to recover from a stack overflow, for
3564 languages where such a recovery is supported. The default value of 75 words
3565 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
3566 8192 bytes with other exception handling mechanisms should be adequate for
3570 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3571 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3572 in the opposite case.
3574 @defmac STACK_CHECK_MAX_FRAME_SIZE
3575 The maximum size of a stack frame, in bytes. GCC will generate probe
3576 instructions in non-leaf functions to ensure at least this many bytes of
3577 stack are available. If a stack frame is larger than this size, stack
3578 checking will not be reliable and GCC will issue a warning. The
3579 default is chosen so that GCC only generates one instruction on most
3580 systems. You should normally not change the default value of this macro.
3583 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3584 GCC uses this value to generate the above warning message. It
3585 represents the amount of fixed frame used by a function, not including
3586 space for any callee-saved registers, temporaries and user variables.
3587 You need only specify an upper bound for this amount and will normally
3588 use the default of four words.
3591 @defmac STACK_CHECK_MAX_VAR_SIZE
3592 The maximum size, in bytes, of an object that GCC will place in the
3593 fixed area of the stack frame when the user specifies
3594 @option{-fstack-check}.
3595 GCC computed the default from the values of the above macros and you will
3596 normally not need to override that default.
3600 @node Frame Registers
3601 @subsection Registers That Address the Stack Frame
3603 @c prevent bad page break with this line
3604 This discusses registers that address the stack frame.
3606 @defmac STACK_POINTER_REGNUM
3607 The register number of the stack pointer register, which must also be a
3608 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3609 the hardware determines which register this is.
3612 @defmac FRAME_POINTER_REGNUM
3613 The register number of the frame pointer register, which is used to
3614 access automatic variables in the stack frame. On some machines, the
3615 hardware determines which register this is. On other machines, you can
3616 choose any register you wish for this purpose.
3619 @defmac HARD_FRAME_POINTER_REGNUM
3620 On some machines the offset between the frame pointer and starting
3621 offset of the automatic variables is not known until after register
3622 allocation has been done (for example, because the saved registers are
3623 between these two locations). On those machines, define
3624 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3625 be used internally until the offset is known, and define
3626 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3627 used for the frame pointer.
3629 You should define this macro only in the very rare circumstances when it
3630 is not possible to calculate the offset between the frame pointer and
3631 the automatic variables until after register allocation has been
3632 completed. When this macro is defined, you must also indicate in your
3633 definition of @code{ELIMINABLE_REGS} how to eliminate
3634 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3635 or @code{STACK_POINTER_REGNUM}.
3637 Do not define this macro if it would be the same as
3638 @code{FRAME_POINTER_REGNUM}.
3641 @defmac ARG_POINTER_REGNUM
3642 The register number of the arg pointer register, which is used to access
3643 the function's argument list. On some machines, this is the same as the
3644 frame pointer register. On some machines, the hardware determines which
3645 register this is. On other machines, you can choose any register you
3646 wish for this purpose. If this is not the same register as the frame
3647 pointer register, then you must mark it as a fixed register according to
3648 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3649 (@pxref{Elimination}).
3652 @defmac RETURN_ADDRESS_POINTER_REGNUM
3653 The register number of the return address pointer register, which is used to
3654 access the current function's return address from the stack. On some
3655 machines, the return address is not at a fixed offset from the frame
3656 pointer or stack pointer or argument pointer. This register can be defined
3657 to point to the return address on the stack, and then be converted by
3658 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3660 Do not define this macro unless there is no other way to get the return
3661 address from the stack.
3664 @defmac STATIC_CHAIN_REGNUM
3665 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3666 Register numbers used for passing a function's static chain pointer. If
3667 register windows are used, the register number as seen by the called
3668 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3669 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3670 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3673 The static chain register need not be a fixed register.
3675 If the static chain is passed in memory, these macros should not be
3676 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3679 @hook TARGET_STATIC_CHAIN
3680 This hook replaces the use of @code{STATIC_CHAIN_REGNUM} et al for
3681 targets that may use different static chain locations for different
3682 nested functions. This may be required if the target has function
3683 attributes that affect the calling conventions of the function and
3684 those calling conventions use different static chain locations.
3686 The default version of this hook uses @code{STATIC_CHAIN_REGNUM} et al.
3688 If the static chain is passed in memory, this hook should be used to
3689 provide rtx giving @code{mem} expressions that denote where they are stored.
3690 Often the @code{mem} expression as seen by the caller will be at an offset
3691 from the stack pointer and the @code{mem} expression as seen by the callee
3692 will be at an offset from the frame pointer.
3693 @findex stack_pointer_rtx
3694 @findex frame_pointer_rtx
3695 @findex arg_pointer_rtx
3696 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3697 @code{arg_pointer_rtx} will have been initialized and should be used
3698 to refer to those items.
3701 @defmac DWARF_FRAME_REGISTERS
3702 This macro specifies the maximum number of hard registers that can be
3703 saved in a call frame. This is used to size data structures used in
3704 DWARF2 exception handling.
3706 Prior to GCC 3.0, this macro was needed in order to establish a stable
3707 exception handling ABI in the face of adding new hard registers for ISA
3708 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3709 in the number of hard registers. Nevertheless, this macro can still be
3710 used to reduce the runtime memory requirements of the exception handling
3711 routines, which can be substantial if the ISA contains a lot of
3712 registers that are not call-saved.
3714 If this macro is not defined, it defaults to
3715 @code{FIRST_PSEUDO_REGISTER}.
3718 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3720 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3721 for backward compatibility in pre GCC 3.0 compiled code.
3723 If this macro is not defined, it defaults to
3724 @code{DWARF_FRAME_REGISTERS}.
3727 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3729 Define this macro if the target's representation for dwarf registers
3730 is different than the internal representation for unwind column.
3731 Given a dwarf register, this macro should return the internal unwind
3732 column number to use instead.
3734 See the PowerPC's SPE target for an example.
3737 @defmac DWARF_FRAME_REGNUM (@var{regno})
3739 Define this macro if the target's representation for dwarf registers
3740 used in .eh_frame or .debug_frame is different from that used in other
3741 debug info sections. Given a GCC hard register number, this macro
3742 should return the .eh_frame register number. The default is
3743 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3747 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3749 Define this macro to map register numbers held in the call frame info
3750 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3751 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3752 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3753 return @code{@var{regno}}.
3758 @subsection Eliminating Frame Pointer and Arg Pointer
3760 @c prevent bad page break with this line
3761 This is about eliminating the frame pointer and arg pointer.
3763 @hook TARGET_FRAME_POINTER_REQUIRED
3764 This target hook should return @code{true} if a function must have and use
3765 a frame pointer. This target hook is called in the reload pass. If its return
3766 value is @code{true} the function will have a frame pointer.
3768 This target hook can in principle examine the current function and decide
3769 according to the facts, but on most machines the constant @code{false} or the
3770 constant @code{true} suffices. Use @code{false} when the machine allows code
3771 to be generated with no frame pointer, and doing so saves some time or space.
3772 Use @code{true} when there is no possible advantage to avoiding a frame
3775 In certain cases, the compiler does not know how to produce valid code
3776 without a frame pointer. The compiler recognizes those cases and
3777 automatically gives the function a frame pointer regardless of what
3778 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3781 In a function that does not require a frame pointer, the frame pointer
3782 register can be allocated for ordinary usage, unless you mark it as a
3783 fixed register. See @code{FIXED_REGISTERS} for more information.
3785 Default return value is @code{false}.
3788 @findex get_frame_size
3789 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3790 A C statement to store in the variable @var{depth-var} the difference
3791 between the frame pointer and the stack pointer values immediately after
3792 the function prologue. The value would be computed from information
3793 such as the result of @code{get_frame_size ()} and the tables of
3794 registers @code{regs_ever_live} and @code{call_used_regs}.
3796 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3797 need not be defined. Otherwise, it must be defined even if
3798 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3799 case, you may set @var{depth-var} to anything.
3802 @defmac ELIMINABLE_REGS
3803 If defined, this macro specifies a table of register pairs used to
3804 eliminate unneeded registers that point into the stack frame. If it is not
3805 defined, the only elimination attempted by the compiler is to replace
3806 references to the frame pointer with references to the stack pointer.
3808 The definition of this macro is a list of structure initializations, each
3809 of which specifies an original and replacement register.
3811 On some machines, the position of the argument pointer is not known until
3812 the compilation is completed. In such a case, a separate hard register
3813 must be used for the argument pointer. This register can be eliminated by
3814 replacing it with either the frame pointer or the argument pointer,
3815 depending on whether or not the frame pointer has been eliminated.
3817 In this case, you might specify:
3819 #define ELIMINABLE_REGS \
3820 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3821 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3822 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3825 Note that the elimination of the argument pointer with the stack pointer is
3826 specified first since that is the preferred elimination.
3829 @hook TARGET_CAN_ELIMINATE
3830 This target hook should returns @code{true} if the compiler is allowed to
3831 try to replace register number @var{from_reg} with register number
3832 @var{to_reg}. This target hook need only be defined if @code{ELIMINABLE_REGS}
3833 is defined, and will usually be @code{true}, since most of the cases
3834 preventing register elimination are things that the compiler already
3837 Default return value is @code{true}.
3840 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3841 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3842 specifies the initial difference between the specified pair of
3843 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3847 @node Stack Arguments
3848 @subsection Passing Function Arguments on the Stack
3849 @cindex arguments on stack
3850 @cindex stack arguments
3852 The macros in this section control how arguments are passed
3853 on the stack. See the following section for other macros that
3854 control passing certain arguments in registers.
3856 @hook TARGET_PROMOTE_PROTOTYPES
3857 This target hook returns @code{true} if an argument declared in a
3858 prototype as an integral type smaller than @code{int} should actually be
3859 passed as an @code{int}. In addition to avoiding errors in certain
3860 cases of mismatch, it also makes for better code on certain machines.
3861 The default is to not promote prototypes.
3865 A C expression. If nonzero, push insns will be used to pass
3867 If the target machine does not have a push instruction, set it to zero.
3868 That directs GCC to use an alternate strategy: to
3869 allocate the entire argument block and then store the arguments into
3870 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3873 @defmac PUSH_ARGS_REVERSED
3874 A C expression. If nonzero, function arguments will be evaluated from
3875 last to first, rather than from first to last. If this macro is not
3876 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3877 and args grow in opposite directions, and 0 otherwise.
3880 @defmac PUSH_ROUNDING (@var{npushed})
3881 A C expression that is the number of bytes actually pushed onto the
3882 stack when an instruction attempts to push @var{npushed} bytes.
3884 On some machines, the definition
3887 #define PUSH_ROUNDING(BYTES) (BYTES)
3891 will suffice. But on other machines, instructions that appear
3892 to push one byte actually push two bytes in an attempt to maintain
3893 alignment. Then the definition should be
3896 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3900 @findex current_function_outgoing_args_size
3901 @defmac ACCUMULATE_OUTGOING_ARGS
3902 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3903 will be computed and placed into the variable
3904 @code{current_function_outgoing_args_size}. No space will be pushed
3905 onto the stack for each call; instead, the function prologue should
3906 increase the stack frame size by this amount.
3908 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3912 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3913 Define this macro if functions should assume that stack space has been
3914 allocated for arguments even when their values are passed in
3917 The value of this macro is the size, in bytes, of the area reserved for
3918 arguments passed in registers for the function represented by @var{fndecl},
3919 which can be zero if GCC is calling a library function.
3920 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3923 This space can be allocated by the caller, or be a part of the
3924 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3927 @c above is overfull. not sure what to do. --mew 5feb93 did
3928 @c something, not sure if it looks good. --mew 10feb93
3930 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3931 Define this to a nonzero value if it is the responsibility of the
3932 caller to allocate the area reserved for arguments passed in registers
3933 when calling a function of @var{fntype}. @var{fntype} may be NULL
3934 if the function called is a library function.
3936 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3937 whether the space for these arguments counts in the value of
3938 @code{current_function_outgoing_args_size}.
3941 @defmac STACK_PARMS_IN_REG_PARM_AREA
3942 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3943 stack parameters don't skip the area specified by it.
3944 @c i changed this, makes more sens and it should have taken care of the
3945 @c overfull.. not as specific, tho. --mew 5feb93
3947 Normally, when a parameter is not passed in registers, it is placed on the
3948 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3949 suppresses this behavior and causes the parameter to be passed on the
3950 stack in its natural location.
3953 @hook TARGET_RETURN_POPS_ARGS
3954 This target hook returns the number of bytes of its own arguments that
3955 a function pops on returning, or 0 if the function pops no arguments
3956 and the caller must therefore pop them all after the function returns.
3958 @var{fundecl} is a C variable whose value is a tree node that describes
3959 the function in question. Normally it is a node of type
3960 @code{FUNCTION_DECL} that describes the declaration of the function.
3961 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3963 @var{funtype} is a C variable whose value is a tree node that
3964 describes the function in question. Normally it is a node of type
3965 @code{FUNCTION_TYPE} that describes the data type of the function.
3966 From this it is possible to obtain the data types of the value and
3967 arguments (if known).
3969 When a call to a library function is being considered, @var{fundecl}
3970 will contain an identifier node for the library function. Thus, if
3971 you need to distinguish among various library functions, you can do so
3972 by their names. Note that ``library function'' in this context means
3973 a function used to perform arithmetic, whose name is known specially
3974 in the compiler and was not mentioned in the C code being compiled.
3976 @var{size} is the number of bytes of arguments passed on the
3977 stack. If a variable number of bytes is passed, it is zero, and
3978 argument popping will always be the responsibility of the calling function.
3980 On the VAX, all functions always pop their arguments, so the definition
3981 of this macro is @var{size}. On the 68000, using the standard
3982 calling convention, no functions pop their arguments, so the value of
3983 the macro is always 0 in this case. But an alternative calling
3984 convention is available in which functions that take a fixed number of
3985 arguments pop them but other functions (such as @code{printf}) pop
3986 nothing (the caller pops all). When this convention is in use,
3987 @var{funtype} is examined to determine whether a function takes a fixed
3988 number of arguments.
3991 @defmac CALL_POPS_ARGS (@var{cum})
3992 A C expression that should indicate the number of bytes a call sequence
3993 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3994 when compiling a function call.
3996 @var{cum} is the variable in which all arguments to the called function
3997 have been accumulated.
3999 On certain architectures, such as the SH5, a call trampoline is used
4000 that pops certain registers off the stack, depending on the arguments
4001 that have been passed to the function. Since this is a property of the
4002 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
4006 @node Register Arguments
4007 @subsection Passing Arguments in Registers
4008 @cindex arguments in registers
4009 @cindex registers arguments
4011 This section describes the macros which let you control how various
4012 types of arguments are passed in registers or how they are arranged in
4015 @defmac FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
4016 A C expression that controls whether a function argument is passed
4017 in a register, and which register.
4019 The arguments are @var{cum}, which summarizes all the previous
4020 arguments; @var{mode}, the machine mode of the argument; @var{type},
4021 the data type of the argument as a tree node or 0 if that is not known
4022 (which happens for C support library functions); and @var{named},
4023 which is 1 for an ordinary argument and 0 for nameless arguments that
4024 correspond to @samp{@dots{}} in the called function's prototype.
4025 @var{type} can be an incomplete type if a syntax error has previously
4028 The value of the expression is usually either a @code{reg} RTX for the
4029 hard register in which to pass the argument, or zero to pass the
4030 argument on the stack.
4032 For machines like the VAX and 68000, where normally all arguments are
4033 pushed, zero suffices as a definition.
4035 The value of the expression can also be a @code{parallel} RTX@. This is
4036 used when an argument is passed in multiple locations. The mode of the
4037 @code{parallel} should be the mode of the entire argument. The
4038 @code{parallel} holds any number of @code{expr_list} pairs; each one
4039 describes where part of the argument is passed. In each
4040 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4041 register in which to pass this part of the argument, and the mode of the
4042 register RTX indicates how large this part of the argument is. The
4043 second operand of the @code{expr_list} is a @code{const_int} which gives
4044 the offset in bytes into the entire argument of where this part starts.
4045 As a special exception the first @code{expr_list} in the @code{parallel}
4046 RTX may have a first operand of zero. This indicates that the entire
4047 argument is also stored on the stack.
4049 The last time this macro is called, it is called with @code{MODE ==
4050 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4051 pattern as operands 2 and 3 respectively.
4053 @cindex @file{stdarg.h} and register arguments
4054 The usual way to make the ISO library @file{stdarg.h} work on a machine
4055 where some arguments are usually passed in registers, is to cause
4056 nameless arguments to be passed on the stack instead. This is done
4057 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
4059 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
4060 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
4061 You may use the hook @code{targetm.calls.must_pass_in_stack}
4062 in the definition of this macro to determine if this argument is of a
4063 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4064 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
4065 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4066 defined, the argument will be computed in the stack and then loaded into
4070 @hook TARGET_MUST_PASS_IN_STACK
4071 This target hook should return @code{true} if we should not pass @var{type}
4072 solely in registers. The file @file{expr.h} defines a
4073 definition that is usually appropriate, refer to @file{expr.h} for additional
4077 @defmac FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
4078 Define this macro if the target machine has ``register windows'', so
4079 that the register in which a function sees an arguments is not
4080 necessarily the same as the one in which the caller passed the
4083 For such machines, @code{FUNCTION_ARG} computes the register in which
4084 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
4085 be defined in a similar fashion to tell the function being called
4086 where the arguments will arrive.
4088 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
4089 serves both purposes.
4092 @hook TARGET_ARG_PARTIAL_BYTES
4093 This target hook returns the number of bytes at the beginning of an
4094 argument that must be put in registers. The value must be zero for
4095 arguments that are passed entirely in registers or that are entirely
4096 pushed on the stack.
4098 On some machines, certain arguments must be passed partially in
4099 registers and partially in memory. On these machines, typically the
4100 first few words of arguments are passed in registers, and the rest
4101 on the stack. If a multi-word argument (a @code{double} or a
4102 structure) crosses that boundary, its first few words must be passed
4103 in registers and the rest must be pushed. This macro tells the
4104 compiler when this occurs, and how many bytes should go in registers.
4106 @code{FUNCTION_ARG} for these arguments should return the first
4107 register to be used by the caller for this argument; likewise
4108 @code{FUNCTION_INCOMING_ARG}, for the called function.
4111 @hook TARGET_PASS_BY_REFERENCE
4112 This target hook should return @code{true} if an argument at the
4113 position indicated by @var{cum} should be passed by reference. This
4114 predicate is queried after target independent reasons for being
4115 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4117 If the hook returns true, a copy of that argument is made in memory and a
4118 pointer to the argument is passed instead of the argument itself.
4119 The pointer is passed in whatever way is appropriate for passing a pointer
4123 @hook TARGET_CALLEE_COPIES
4124 The function argument described by the parameters to this hook is
4125 known to be passed by reference. The hook should return true if the
4126 function argument should be copied by the callee instead of copied
4129 For any argument for which the hook returns true, if it can be
4130 determined that the argument is not modified, then a copy need
4133 The default version of this hook always returns false.
4136 @defmac CUMULATIVE_ARGS
4137 A C type for declaring a variable that is used as the first argument of
4138 @code{FUNCTION_ARG} and other related values. For some target machines,
4139 the type @code{int} suffices and can hold the number of bytes of
4142 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4143 arguments that have been passed on the stack. The compiler has other
4144 variables to keep track of that. For target machines on which all
4145 arguments are passed on the stack, there is no need to store anything in
4146 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4147 should not be empty, so use @code{int}.
4150 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4151 If defined, this macro is called before generating any code for a
4152 function, but after the @var{cfun} descriptor for the function has been
4153 created. The back end may use this macro to update @var{cfun} to
4154 reflect an ABI other than that which would normally be used by default.
4155 If the compiler is generating code for a compiler-generated function,
4156 @var{fndecl} may be @code{NULL}.
4159 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4160 A C statement (sans semicolon) for initializing the variable
4161 @var{cum} for the state at the beginning of the argument list. The
4162 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4163 is the tree node for the data type of the function which will receive
4164 the args, or 0 if the args are to a compiler support library function.
4165 For direct calls that are not libcalls, @var{fndecl} contain the
4166 declaration node of the function. @var{fndecl} is also set when
4167 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4168 being compiled. @var{n_named_args} is set to the number of named
4169 arguments, including a structure return address if it is passed as a
4170 parameter, when making a call. When processing incoming arguments,
4171 @var{n_named_args} is set to @minus{}1.
4173 When processing a call to a compiler support library function,
4174 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4175 contains the name of the function, as a string. @var{libname} is 0 when
4176 an ordinary C function call is being processed. Thus, each time this
4177 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4178 never both of them at once.
4181 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4182 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4183 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4184 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4185 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4186 0)} is used instead.
4189 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4190 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4191 finding the arguments for the function being compiled. If this macro is
4192 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4194 The value passed for @var{libname} is always 0, since library routines
4195 with special calling conventions are never compiled with GCC@. The
4196 argument @var{libname} exists for symmetry with
4197 @code{INIT_CUMULATIVE_ARGS}.
4198 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4199 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4202 @defmac FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
4203 A C statement (sans semicolon) to update the summarizer variable
4204 @var{cum} to advance past an argument in the argument list. The
4205 values @var{mode}, @var{type} and @var{named} describe that argument.
4206 Once this is done, the variable @var{cum} is suitable for analyzing
4207 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
4209 This macro need not do anything if the argument in question was passed
4210 on the stack. The compiler knows how to track the amount of stack space
4211 used for arguments without any special help.
4214 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4215 If defined, a C expression that is the number of bytes to add to the
4216 offset of the argument passed in memory. This is needed for the SPU,
4217 which passes @code{char} and @code{short} arguments in the preferred
4218 slot that is in the middle of the quad word instead of starting at the
4222 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4223 If defined, a C expression which determines whether, and in which direction,
4224 to pad out an argument with extra space. The value should be of type
4225 @code{enum direction}: either @code{upward} to pad above the argument,
4226 @code{downward} to pad below, or @code{none} to inhibit padding.
4228 The @emph{amount} of padding is always just enough to reach the next
4229 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
4232 This macro has a default definition which is right for most systems.
4233 For little-endian machines, the default is to pad upward. For
4234 big-endian machines, the default is to pad downward for an argument of
4235 constant size shorter than an @code{int}, and upward otherwise.
4238 @defmac PAD_VARARGS_DOWN
4239 If defined, a C expression which determines whether the default
4240 implementation of va_arg will attempt to pad down before reading the
4241 next argument, if that argument is smaller than its aligned space as
4242 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4243 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4246 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4247 Specify padding for the last element of a block move between registers and
4248 memory. @var{first} is nonzero if this is the only element. Defining this
4249 macro allows better control of register function parameters on big-endian
4250 machines, without using @code{PARALLEL} rtl. In particular,
4251 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4252 registers, as there is no longer a "wrong" part of a register; For example,
4253 a three byte aggregate may be passed in the high part of a register if so
4257 @defmac FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
4258 If defined, a C expression that gives the alignment boundary, in bits,
4259 of an argument with the specified mode and type. If it is not defined,
4260 @code{PARM_BOUNDARY} is used for all arguments.
4263 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4264 A C expression that is nonzero if @var{regno} is the number of a hard
4265 register in which function arguments are sometimes passed. This does
4266 @emph{not} include implicit arguments such as the static chain and
4267 the structure-value address. On many machines, no registers can be
4268 used for this purpose since all function arguments are pushed on the
4272 @hook TARGET_SPLIT_COMPLEX_ARG
4273 This hook should return true if parameter of type @var{type} are passed
4274 as two scalar parameters. By default, GCC will attempt to pack complex
4275 arguments into the target's word size. Some ABIs require complex arguments
4276 to be split and treated as their individual components. For example, on
4277 AIX64, complex floats should be passed in a pair of floating point
4278 registers, even though a complex float would fit in one 64-bit floating
4281 The default value of this hook is @code{NULL}, which is treated as always
4285 @hook TARGET_BUILD_BUILTIN_VA_LIST
4286 This hook returns a type node for @code{va_list} for the target.
4287 The default version of the hook returns @code{void*}.
4290 @hook TARGET_ENUM_VA_LIST_P
4291 This target hook is used in function @code{c_common_nodes_and_builtins}
4292 to iterate through the target specific builtin types for va_list. The
4293 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4294 to a @code{const char *} and @var{ptree} a pointer to a @code{tree} typed
4296 The arguments @var{pname} and @var{ptree} are used to store the result of
4297 this macro and are set to the name of the va_list builtin type and its
4299 If the return value of this macro is zero, then there is no more element.
4300 Otherwise the @var{IDX} should be increased for the next call of this
4301 macro to iterate through all types.
4304 @hook TARGET_FN_ABI_VA_LIST
4305 This hook returns the va_list type of the calling convention specified by
4307 The default version of this hook returns @code{va_list_type_node}.
4310 @hook TARGET_CANONICAL_VA_LIST_TYPE
4311 This hook returns the va_list type of the calling convention specified by the
4312 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4316 @hook TARGET_GIMPLIFY_VA_ARG_EXPR
4317 This hook performs target-specific gimplification of
4318 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4319 arguments to @code{va_arg}; the latter two are as in
4320 @code{gimplify.c:gimplify_expr}.
4323 @hook TARGET_VALID_POINTER_MODE
4324 Define this to return nonzero if the port can handle pointers
4325 with machine mode @var{mode}. The default version of this
4326 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4329 @hook TARGET_SCALAR_MODE_SUPPORTED_P
4330 Define this to return nonzero if the port is prepared to handle
4331 insns involving scalar mode @var{mode}. For a scalar mode to be
4332 considered supported, all the basic arithmetic and comparisons
4335 The default version of this hook returns true for any mode
4336 required to handle the basic C types (as defined by the port).
4337 Included here are the double-word arithmetic supported by the
4338 code in @file{optabs.c}.
4341 @hook TARGET_VECTOR_MODE_SUPPORTED_P
4342 Define this to return nonzero if the port is prepared to handle
4343 insns involving vector mode @var{mode}. At the very least, it
4344 must have move patterns for this mode.
4347 @hook TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P
4348 Define this to return nonzero for machine modes for which the port has
4349 small register classes. If this target hook returns nonzero for a given
4350 @var{mode}, the compiler will try to minimize the lifetime of registers
4351 in @var{mode}. The hook may be called with @code{VOIDmode} as argument.
4352 In this case, the hook is expected to return nonzero if it returns nonzero
4355 On some machines, it is risky to let hard registers live across arbitrary
4356 insns. Typically, these machines have instructions that require values
4357 to be in specific registers (like an accumulator), and reload will fail
4358 if the required hard register is used for another purpose across such an
4361 Passes before reload do not know which hard registers will be used
4362 in an instruction, but the machine modes of the registers set or used in
4363 the instruction are already known. And for some machines, register
4364 classes are small for, say, integer registers but not for floating point
4365 registers. For example, the AMD x86-64 architecture requires specific
4366 registers for the legacy x86 integer instructions, but there are many
4367 SSE registers for floating point operations. On such targets, a good
4368 strategy may be to return nonzero from this hook for @code{INTEGRAL_MODE_P}
4369 machine modes but zero for the SSE register classes.
4371 The default version of this hook retuns false for any mode. It is always
4372 safe to redefine this hook to return with a nonzero value. But if you
4373 unnecessarily define it, you will reduce the amount of optimizations
4374 that can be performed in some cases. If you do not define this hook
4375 to return a nonzero value when it is required, the compiler will run out
4376 of spill registers and print a fatal error message.
4380 @subsection How Scalar Function Values Are Returned
4381 @cindex return values in registers
4382 @cindex values, returned by functions
4383 @cindex scalars, returned as values
4385 This section discusses the macros that control returning scalars as
4386 values---values that can fit in registers.
4388 @hook TARGET_FUNCTION_VALUE
4390 Define this to return an RTX representing the place where a function
4391 returns or receives a value of data type @var{ret_type}, a tree node
4392 representing a data type. @var{fn_decl_or_type} is a tree node
4393 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4394 function being called. If @var{outgoing} is false, the hook should
4395 compute the register in which the caller will see the return value.
4396 Otherwise, the hook should return an RTX representing the place where
4397 a function returns a value.
4399 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4400 (Actually, on most machines, scalar values are returned in the same
4401 place regardless of mode.) The value of the expression is usually a
4402 @code{reg} RTX for the hard register where the return value is stored.
4403 The value can also be a @code{parallel} RTX, if the return value is in
4404 multiple places. See @code{FUNCTION_ARG} for an explanation of the
4405 @code{parallel} form. Note that the callee will populate every
4406 location specified in the @code{parallel}, but if the first element of
4407 the @code{parallel} contains the whole return value, callers will use
4408 that element as the canonical location and ignore the others. The m68k
4409 port uses this type of @code{parallel} to return pointers in both
4410 @samp{%a0} (the canonical location) and @samp{%d0}.
4412 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4413 the same promotion rules specified in @code{PROMOTE_MODE} if
4414 @var{valtype} is a scalar type.
4416 If the precise function being called is known, @var{func} is a tree
4417 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4418 pointer. This makes it possible to use a different value-returning
4419 convention for specific functions when all their calls are
4422 Some target machines have ``register windows'' so that the register in
4423 which a function returns its value is not the same as the one in which
4424 the caller sees the value. For such machines, you should return
4425 different RTX depending on @var{outgoing}.
4427 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4428 aggregate data types, because these are returned in another way. See
4429 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4432 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4433 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4434 a new target instead.
4437 @defmac LIBCALL_VALUE (@var{mode})
4438 A C expression to create an RTX representing the place where a library
4439 function returns a value of mode @var{mode}.
4441 Note that ``library function'' in this context means a compiler
4442 support routine, used to perform arithmetic, whose name is known
4443 specially by the compiler and was not mentioned in the C code being
4447 @hook TARGET_LIBCALL_VALUE
4448 Define this hook if the back-end needs to know the name of the libcall
4449 function in order to determine where the result should be returned.
4451 The mode of the result is given by @var{mode} and the name of the called
4452 library function is given by @var{fun}. The hook should return an RTX
4453 representing the place where the library function result will be returned.
4455 If this hook is not defined, then LIBCALL_VALUE will be used.
4458 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4459 A C expression that is nonzero if @var{regno} is the number of a hard
4460 register in which the values of called function may come back.
4462 A register whose use for returning values is limited to serving as the
4463 second of a pair (for a value of type @code{double}, say) need not be
4464 recognized by this macro. So for most machines, this definition
4468 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4471 If the machine has register windows, so that the caller and the called
4472 function use different registers for the return value, this macro
4473 should recognize only the caller's register numbers.
4475 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
4476 for a new target instead.
4479 @hook TARGET_FUNCTION_VALUE_REGNO_P
4480 A target hook that return @code{true} if @var{regno} is the number of a hard
4481 register in which the values of called function may come back.
4483 A register whose use for returning values is limited to serving as the
4484 second of a pair (for a value of type @code{double}, say) need not be
4485 recognized by this target hook.
4487 If the machine has register windows, so that the caller and the called
4488 function use different registers for the return value, this target hook
4489 should recognize only the caller's register numbers.
4491 If this hook is not defined, then FUNCTION_VALUE_REGNO_P will be used.
4494 @defmac APPLY_RESULT_SIZE
4495 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4496 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4497 saving and restoring an arbitrary return value.
4500 @hook TARGET_RETURN_IN_MSB
4501 This hook should return true if values of type @var{type} are returned
4502 at the most significant end of a register (in other words, if they are
4503 padded at the least significant end). You can assume that @var{type}
4504 is returned in a register; the caller is required to check this.
4506 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4507 be able to hold the complete return value. For example, if a 1-, 2-
4508 or 3-byte structure is returned at the most significant end of a
4509 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4513 @node Aggregate Return
4514 @subsection How Large Values Are Returned
4515 @cindex aggregates as return values
4516 @cindex large return values
4517 @cindex returning aggregate values
4518 @cindex structure value address
4520 When a function value's mode is @code{BLKmode} (and in some other
4521 cases), the value is not returned according to
4522 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4523 caller passes the address of a block of memory in which the value
4524 should be stored. This address is called the @dfn{structure value
4527 This section describes how to control returning structure values in
4530 @hook TARGET_RETURN_IN_MEMORY
4531 This target hook should return a nonzero value to say to return the
4532 function value in memory, just as large structures are always returned.
4533 Here @var{type} will be the data type of the value, and @var{fntype}
4534 will be the type of the function doing the returning, or @code{NULL} for
4537 Note that values of mode @code{BLKmode} must be explicitly handled
4538 by this function. Also, the option @option{-fpcc-struct-return}
4539 takes effect regardless of this macro. On most systems, it is
4540 possible to leave the hook undefined; this causes a default
4541 definition to be used, whose value is the constant 1 for @code{BLKmode}
4542 values, and 0 otherwise.
4544 Do not use this hook to indicate that structures and unions should always
4545 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4549 @defmac DEFAULT_PCC_STRUCT_RETURN
4550 Define this macro to be 1 if all structure and union return values must be
4551 in memory. Since this results in slower code, this should be defined
4552 only if needed for compatibility with other compilers or with an ABI@.
4553 If you define this macro to be 0, then the conventions used for structure
4554 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4557 If not defined, this defaults to the value 1.
4560 @hook TARGET_STRUCT_VALUE_RTX
4561 This target hook should return the location of the structure value
4562 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4563 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4564 be @code{NULL}, for libcalls. You do not need to define this target
4565 hook if the address is always passed as an ``invisible'' first
4568 On some architectures the place where the structure value address
4569 is found by the called function is not the same place that the
4570 caller put it. This can be due to register windows, or it could
4571 be because the function prologue moves it to a different place.
4572 @var{incoming} is @code{1} or @code{2} when the location is needed in
4573 the context of the called function, and @code{0} in the context of
4576 If @var{incoming} is nonzero and the address is to be found on the
4577 stack, return a @code{mem} which refers to the frame pointer. If
4578 @var{incoming} is @code{2}, the result is being used to fetch the
4579 structure value address at the beginning of a function. If you need
4580 to emit adjusting code, you should do it at this point.
4583 @defmac PCC_STATIC_STRUCT_RETURN
4584 Define this macro if the usual system convention on the target machine
4585 for returning structures and unions is for the called function to return
4586 the address of a static variable containing the value.
4588 Do not define this if the usual system convention is for the caller to
4589 pass an address to the subroutine.
4591 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4592 nothing when you use @option{-freg-struct-return} mode.
4596 @subsection Caller-Saves Register Allocation
4598 If you enable it, GCC can save registers around function calls. This
4599 makes it possible to use call-clobbered registers to hold variables that
4600 must live across calls.
4602 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4603 A C expression to determine whether it is worthwhile to consider placing
4604 a pseudo-register in a call-clobbered hard register and saving and
4605 restoring it around each function call. The expression should be 1 when
4606 this is worth doing, and 0 otherwise.
4608 If you don't define this macro, a default is used which is good on most
4609 machines: @code{4 * @var{calls} < @var{refs}}.
4612 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4613 A C expression specifying which mode is required for saving @var{nregs}
4614 of a pseudo-register in call-clobbered hard register @var{regno}. If
4615 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4616 returned. For most machines this macro need not be defined since GCC
4617 will select the smallest suitable mode.
4620 @node Function Entry
4621 @subsection Function Entry and Exit
4622 @cindex function entry and exit
4626 This section describes the macros that output function entry
4627 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4629 @hook TARGET_ASM_FUNCTION_PROLOGUE
4630 If defined, a function that outputs the assembler code for entry to a
4631 function. The prologue is responsible for setting up the stack frame,
4632 initializing the frame pointer register, saving registers that must be
4633 saved, and allocating @var{size} additional bytes of storage for the
4634 local variables. @var{size} is an integer. @var{file} is a stdio
4635 stream to which the assembler code should be output.
4637 The label for the beginning of the function need not be output by this
4638 macro. That has already been done when the macro is run.
4640 @findex regs_ever_live
4641 To determine which registers to save, the macro can refer to the array
4642 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4643 @var{r} is used anywhere within the function. This implies the function
4644 prologue should save register @var{r}, provided it is not one of the
4645 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4646 @code{regs_ever_live}.)
4648 On machines that have ``register windows'', the function entry code does
4649 not save on the stack the registers that are in the windows, even if
4650 they are supposed to be preserved by function calls; instead it takes
4651 appropriate steps to ``push'' the register stack, if any non-call-used
4652 registers are used in the function.
4654 @findex frame_pointer_needed
4655 On machines where functions may or may not have frame-pointers, the
4656 function entry code must vary accordingly; it must set up the frame
4657 pointer if one is wanted, and not otherwise. To determine whether a
4658 frame pointer is in wanted, the macro can refer to the variable
4659 @code{frame_pointer_needed}. The variable's value will be 1 at run
4660 time in a function that needs a frame pointer. @xref{Elimination}.
4662 The function entry code is responsible for allocating any stack space
4663 required for the function. This stack space consists of the regions
4664 listed below. In most cases, these regions are allocated in the
4665 order listed, with the last listed region closest to the top of the
4666 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4667 the highest address if it is not defined). You can use a different order
4668 for a machine if doing so is more convenient or required for
4669 compatibility reasons. Except in cases where required by standard
4670 or by a debugger, there is no reason why the stack layout used by GCC
4671 need agree with that used by other compilers for a machine.
4674 @hook TARGET_ASM_FUNCTION_END_PROLOGUE
4675 If defined, a function that outputs assembler code at the end of a
4676 prologue. This should be used when the function prologue is being
4677 emitted as RTL, and you have some extra assembler that needs to be
4678 emitted. @xref{prologue instruction pattern}.
4681 @hook TARGET_ASM_FUNCTION_BEGIN_EPILOGUE
4682 If defined, a function that outputs assembler code at the start of an
4683 epilogue. This should be used when the function epilogue is being
4684 emitted as RTL, and you have some extra assembler that needs to be
4685 emitted. @xref{epilogue instruction pattern}.
4688 @hook TARGET_ASM_FUNCTION_EPILOGUE
4689 If defined, a function that outputs the assembler code for exit from a
4690 function. The epilogue is responsible for restoring the saved
4691 registers and stack pointer to their values when the function was
4692 called, and returning control to the caller. This macro takes the
4693 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4694 registers to restore are determined from @code{regs_ever_live} and
4695 @code{CALL_USED_REGISTERS} in the same way.
4697 On some machines, there is a single instruction that does all the work
4698 of returning from the function. On these machines, give that
4699 instruction the name @samp{return} and do not define the macro
4700 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4702 Do not define a pattern named @samp{return} if you want the
4703 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4704 switches to control whether return instructions or epilogues are used,
4705 define a @samp{return} pattern with a validity condition that tests the
4706 target switches appropriately. If the @samp{return} pattern's validity
4707 condition is false, epilogues will be used.
4709 On machines where functions may or may not have frame-pointers, the
4710 function exit code must vary accordingly. Sometimes the code for these
4711 two cases is completely different. To determine whether a frame pointer
4712 is wanted, the macro can refer to the variable
4713 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4714 a function that needs a frame pointer.
4716 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4717 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4718 The C variable @code{current_function_is_leaf} is nonzero for such a
4719 function. @xref{Leaf Functions}.
4721 On some machines, some functions pop their arguments on exit while
4722 others leave that for the caller to do. For example, the 68020 when
4723 given @option{-mrtd} pops arguments in functions that take a fixed
4724 number of arguments.
4726 @findex current_function_pops_args
4727 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4728 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4729 needs to know what was decided. The number of bytes of the current
4730 function's arguments that this function should pop is available in
4731 @code{crtl->args.pops_args}. @xref{Scalar Return}.
4736 @findex current_function_pretend_args_size
4737 A region of @code{current_function_pretend_args_size} bytes of
4738 uninitialized space just underneath the first argument arriving on the
4739 stack. (This may not be at the very start of the allocated stack region
4740 if the calling sequence has pushed anything else since pushing the stack
4741 arguments. But usually, on such machines, nothing else has been pushed
4742 yet, because the function prologue itself does all the pushing.) This
4743 region is used on machines where an argument may be passed partly in
4744 registers and partly in memory, and, in some cases to support the
4745 features in @code{<stdarg.h>}.
4748 An area of memory used to save certain registers used by the function.
4749 The size of this area, which may also include space for such things as
4750 the return address and pointers to previous stack frames, is
4751 machine-specific and usually depends on which registers have been used
4752 in the function. Machines with register windows often do not require
4756 A region of at least @var{size} bytes, possibly rounded up to an allocation
4757 boundary, to contain the local variables of the function. On some machines,
4758 this region and the save area may occur in the opposite order, with the
4759 save area closer to the top of the stack.
4762 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4763 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4764 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4765 argument lists of the function. @xref{Stack Arguments}.
4768 @defmac EXIT_IGNORE_STACK
4769 Define this macro as a C expression that is nonzero if the return
4770 instruction or the function epilogue ignores the value of the stack
4771 pointer; in other words, if it is safe to delete an instruction to
4772 adjust the stack pointer before a return from the function. The
4775 Note that this macro's value is relevant only for functions for which
4776 frame pointers are maintained. It is never safe to delete a final
4777 stack adjustment in a function that has no frame pointer, and the
4778 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4781 @defmac EPILOGUE_USES (@var{regno})
4782 Define this macro as a C expression that is nonzero for registers that are
4783 used by the epilogue or the @samp{return} pattern. The stack and frame
4784 pointer registers are already assumed to be used as needed.
4787 @defmac EH_USES (@var{regno})
4788 Define this macro as a C expression that is nonzero for registers that are
4789 used by the exception handling mechanism, and so should be considered live
4790 on entry to an exception edge.
4793 @defmac DELAY_SLOTS_FOR_EPILOGUE
4794 Define this macro if the function epilogue contains delay slots to which
4795 instructions from the rest of the function can be ``moved''. The
4796 definition should be a C expression whose value is an integer
4797 representing the number of delay slots there.
4800 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4801 A C expression that returns 1 if @var{insn} can be placed in delay
4802 slot number @var{n} of the epilogue.
4804 The argument @var{n} is an integer which identifies the delay slot now
4805 being considered (since different slots may have different rules of
4806 eligibility). It is never negative and is always less than the number
4807 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4808 If you reject a particular insn for a given delay slot, in principle, it
4809 may be reconsidered for a subsequent delay slot. Also, other insns may
4810 (at least in principle) be considered for the so far unfilled delay
4813 @findex current_function_epilogue_delay_list
4814 @findex final_scan_insn
4815 The insns accepted to fill the epilogue delay slots are put in an RTL
4816 list made with @code{insn_list} objects, stored in the variable
4817 @code{current_function_epilogue_delay_list}. The insn for the first
4818 delay slot comes first in the list. Your definition of the macro
4819 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4820 outputting the insns in this list, usually by calling
4821 @code{final_scan_insn}.
4823 You need not define this macro if you did not define
4824 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4827 @hook TARGET_ASM_OUTPUT_MI_THUNK
4828 A function that outputs the assembler code for a thunk
4829 function, used to implement C++ virtual function calls with multiple
4830 inheritance. The thunk acts as a wrapper around a virtual function,
4831 adjusting the implicit object parameter before handing control off to
4834 First, emit code to add the integer @var{delta} to the location that
4835 contains the incoming first argument. Assume that this argument
4836 contains a pointer, and is the one used to pass the @code{this} pointer
4837 in C++. This is the incoming argument @emph{before} the function prologue,
4838 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4839 all other incoming arguments.
4841 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4842 made after adding @code{delta}. In particular, if @var{p} is the
4843 adjusted pointer, the following adjustment should be made:
4846 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4849 After the additions, emit code to jump to @var{function}, which is a
4850 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4851 not touch the return address. Hence returning from @var{FUNCTION} will
4852 return to whoever called the current @samp{thunk}.
4854 The effect must be as if @var{function} had been called directly with
4855 the adjusted first argument. This macro is responsible for emitting all
4856 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4857 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4859 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4860 have already been extracted from it.) It might possibly be useful on
4861 some targets, but probably not.
4863 If you do not define this macro, the target-independent code in the C++
4864 front end will generate a less efficient heavyweight thunk that calls
4865 @var{function} instead of jumping to it. The generic approach does
4866 not support varargs.
4869 @hook TARGET_ASM_CAN_OUTPUT_MI_THUNK
4870 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4871 to output the assembler code for the thunk function specified by the
4872 arguments it is passed, and false otherwise. In the latter case, the
4873 generic approach will be used by the C++ front end, with the limitations
4878 @subsection Generating Code for Profiling
4879 @cindex profiling, code generation
4881 These macros will help you generate code for profiling.
4883 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4884 A C statement or compound statement to output to @var{file} some
4885 assembler code to call the profiling subroutine @code{mcount}.
4888 The details of how @code{mcount} expects to be called are determined by
4889 your operating system environment, not by GCC@. To figure them out,
4890 compile a small program for profiling using the system's installed C
4891 compiler and look at the assembler code that results.
4893 Older implementations of @code{mcount} expect the address of a counter
4894 variable to be loaded into some register. The name of this variable is
4895 @samp{LP} followed by the number @var{labelno}, so you would generate
4896 the name using @samp{LP%d} in a @code{fprintf}.
4899 @defmac PROFILE_HOOK
4900 A C statement or compound statement to output to @var{file} some assembly
4901 code to call the profiling subroutine @code{mcount} even the target does
4902 not support profiling.
4905 @defmac NO_PROFILE_COUNTERS
4906 Define this macro to be an expression with a nonzero value if the
4907 @code{mcount} subroutine on your system does not need a counter variable
4908 allocated for each function. This is true for almost all modern
4909 implementations. If you define this macro, you must not use the
4910 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4913 @defmac PROFILE_BEFORE_PROLOGUE
4914 Define this macro if the code for function profiling should come before
4915 the function prologue. Normally, the profiling code comes after.
4919 @subsection Permitting tail calls
4922 @hook TARGET_FUNCTION_OK_FOR_SIBCALL
4923 True if it is ok to do sibling call optimization for the specified
4924 call expression @var{exp}. @var{decl} will be the called function,
4925 or @code{NULL} if this is an indirect call.
4927 It is not uncommon for limitations of calling conventions to prevent
4928 tail calls to functions outside the current unit of translation, or
4929 during PIC compilation. The hook is used to enforce these restrictions,
4930 as the @code{sibcall} md pattern can not fail, or fall over to a
4931 ``normal'' call. The criteria for successful sibling call optimization
4932 may vary greatly between different architectures.
4935 @hook TARGET_EXTRA_LIVE_ON_ENTRY
4936 Add any hard registers to @var{regs} that are live on entry to the
4937 function. This hook only needs to be defined to provide registers that
4938 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4939 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4940 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4941 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4944 @node Stack Smashing Protection
4945 @subsection Stack smashing protection
4946 @cindex stack smashing protection
4948 @hook TARGET_STACK_PROTECT_GUARD
4949 This hook returns a @code{DECL} node for the external variable to use
4950 for the stack protection guard. This variable is initialized by the
4951 runtime to some random value and is used to initialize the guard value
4952 that is placed at the top of the local stack frame. The type of this
4953 variable must be @code{ptr_type_node}.
4955 The default version of this hook creates a variable called
4956 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4959 @hook TARGET_STACK_PROTECT_FAIL
4960 This hook returns a tree expression that alerts the runtime that the
4961 stack protect guard variable has been modified. This expression should
4962 involve a call to a @code{noreturn} function.
4964 The default version of this hook invokes a function called
4965 @samp{__stack_chk_fail}, taking no arguments. This function is
4966 normally defined in @file{libgcc2.c}.
4970 @section Implementing the Varargs Macros
4971 @cindex varargs implementation
4973 GCC comes with an implementation of @code{<varargs.h>} and
4974 @code{<stdarg.h>} that work without change on machines that pass arguments
4975 on the stack. Other machines require their own implementations of
4976 varargs, and the two machine independent header files must have
4977 conditionals to include it.
4979 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4980 the calling convention for @code{va_start}. The traditional
4981 implementation takes just one argument, which is the variable in which
4982 to store the argument pointer. The ISO implementation of
4983 @code{va_start} takes an additional second argument. The user is
4984 supposed to write the last named argument of the function here.
4986 However, @code{va_start} should not use this argument. The way to find
4987 the end of the named arguments is with the built-in functions described
4990 @defmac __builtin_saveregs ()
4991 Use this built-in function to save the argument registers in memory so
4992 that the varargs mechanism can access them. Both ISO and traditional
4993 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4994 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4996 On some machines, @code{__builtin_saveregs} is open-coded under the
4997 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4998 other machines, it calls a routine written in assembler language,
4999 found in @file{libgcc2.c}.
5001 Code generated for the call to @code{__builtin_saveregs} appears at the
5002 beginning of the function, as opposed to where the call to
5003 @code{__builtin_saveregs} is written, regardless of what the code is.
5004 This is because the registers must be saved before the function starts
5005 to use them for its own purposes.
5006 @c i rewrote the first sentence above to fix an overfull hbox. --mew
5010 @defmac __builtin_args_info (@var{category})
5011 Use this built-in function to find the first anonymous arguments in
5014 In general, a machine may have several categories of registers used for
5015 arguments, each for a particular category of data types. (For example,
5016 on some machines, floating-point registers are used for floating-point
5017 arguments while other arguments are passed in the general registers.)
5018 To make non-varargs functions use the proper calling convention, you
5019 have defined the @code{CUMULATIVE_ARGS} data type to record how many
5020 registers in each category have been used so far
5022 @code{__builtin_args_info} accesses the same data structure of type
5023 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
5024 with it, with @var{category} specifying which word to access. Thus, the
5025 value indicates the first unused register in a given category.
5027 Normally, you would use @code{__builtin_args_info} in the implementation
5028 of @code{va_start}, accessing each category just once and storing the
5029 value in the @code{va_list} object. This is because @code{va_list} will
5030 have to update the values, and there is no way to alter the
5031 values accessed by @code{__builtin_args_info}.
5034 @defmac __builtin_next_arg (@var{lastarg})
5035 This is the equivalent of @code{__builtin_args_info}, for stack
5036 arguments. It returns the address of the first anonymous stack
5037 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
5038 returns the address of the location above the first anonymous stack
5039 argument. Use it in @code{va_start} to initialize the pointer for
5040 fetching arguments from the stack. Also use it in @code{va_start} to
5041 verify that the second parameter @var{lastarg} is the last named argument
5042 of the current function.
5045 @defmac __builtin_classify_type (@var{object})
5046 Since each machine has its own conventions for which data types are
5047 passed in which kind of register, your implementation of @code{va_arg}
5048 has to embody these conventions. The easiest way to categorize the
5049 specified data type is to use @code{__builtin_classify_type} together
5050 with @code{sizeof} and @code{__alignof__}.
5052 @code{__builtin_classify_type} ignores the value of @var{object},
5053 considering only its data type. It returns an integer describing what
5054 kind of type that is---integer, floating, pointer, structure, and so on.
5056 The file @file{typeclass.h} defines an enumeration that you can use to
5057 interpret the values of @code{__builtin_classify_type}.
5060 These machine description macros help implement varargs:
5062 @hook TARGET_EXPAND_BUILTIN_SAVEREGS
5063 If defined, this hook produces the machine-specific code for a call to
5064 @code{__builtin_saveregs}. This code will be moved to the very
5065 beginning of the function, before any parameter access are made. The
5066 return value of this function should be an RTX that contains the value
5067 to use as the return of @code{__builtin_saveregs}.
5070 @hook TARGET_SETUP_INCOMING_VARARGS
5071 This target hook offers an alternative to using
5072 @code{__builtin_saveregs} and defining the hook
5073 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
5074 register arguments into the stack so that all the arguments appear to
5075 have been passed consecutively on the stack. Once this is done, you can
5076 use the standard implementation of varargs that works for machines that
5077 pass all their arguments on the stack.
5079 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5080 structure, containing the values that are obtained after processing the
5081 named arguments. The arguments @var{mode} and @var{type} describe the
5082 last named argument---its machine mode and its data type as a tree node.
5084 The target hook should do two things: first, push onto the stack all the
5085 argument registers @emph{not} used for the named arguments, and second,
5086 store the size of the data thus pushed into the @code{int}-valued
5087 variable pointed to by @var{pretend_args_size}. The value that you
5088 store here will serve as additional offset for setting up the stack
5091 Because you must generate code to push the anonymous arguments at
5092 compile time without knowing their data types,
5093 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5094 have just a single category of argument register and use it uniformly
5097 If the argument @var{second_time} is nonzero, it means that the
5098 arguments of the function are being analyzed for the second time. This
5099 happens for an inline function, which is not actually compiled until the
5100 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5101 not generate any instructions in this case.
5104 @hook TARGET_STRICT_ARGUMENT_NAMING
5105 Define this hook to return @code{true} if the location where a function
5106 argument is passed depends on whether or not it is a named argument.
5108 This hook controls how the @var{named} argument to @code{FUNCTION_ARG}
5109 is set for varargs and stdarg functions. If this hook returns
5110 @code{true}, the @var{named} argument is always true for named
5111 arguments, and false for unnamed arguments. If it returns @code{false},
5112 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5113 then all arguments are treated as named. Otherwise, all named arguments
5114 except the last are treated as named.
5116 You need not define this hook if it always returns @code{false}.
5119 @hook TARGET_PRETEND_OUTGOING_VARARGS_NAMED
5120 If you need to conditionally change ABIs so that one works with
5121 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5122 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5123 defined, then define this hook to return @code{true} if
5124 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5125 Otherwise, you should not define this hook.
5129 @section Trampolines for Nested Functions
5130 @cindex trampolines for nested functions
5131 @cindex nested functions, trampolines for
5133 A @dfn{trampoline} is a small piece of code that is created at run time
5134 when the address of a nested function is taken. It normally resides on
5135 the stack, in the stack frame of the containing function. These macros
5136 tell GCC how to generate code to allocate and initialize a
5139 The instructions in the trampoline must do two things: load a constant
5140 address into the static chain register, and jump to the real address of
5141 the nested function. On CISC machines such as the m68k, this requires
5142 two instructions, a move immediate and a jump. Then the two addresses
5143 exist in the trampoline as word-long immediate operands. On RISC
5144 machines, it is often necessary to load each address into a register in
5145 two parts. Then pieces of each address form separate immediate
5148 The code generated to initialize the trampoline must store the variable
5149 parts---the static chain value and the function address---into the
5150 immediate operands of the instructions. On a CISC machine, this is
5151 simply a matter of copying each address to a memory reference at the
5152 proper offset from the start of the trampoline. On a RISC machine, it
5153 may be necessary to take out pieces of the address and store them
5156 @hook TARGET_ASM_TRAMPOLINE_TEMPLATE
5157 This hook is called by @code{assemble_trampoline_template} to output,
5158 on the stream @var{f}, assembler code for a block of data that contains
5159 the constant parts of a trampoline. This code should not include a
5160 label---the label is taken care of automatically.
5162 If you do not define this hook, it means no template is needed
5163 for the target. Do not define this hook on systems where the block move
5164 code to copy the trampoline into place would be larger than the code
5165 to generate it on the spot.
5168 @defmac TRAMPOLINE_SECTION
5169 Return the section into which the trampoline template is to be placed
5170 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5173 @defmac TRAMPOLINE_SIZE
5174 A C expression for the size in bytes of the trampoline, as an integer.
5177 @defmac TRAMPOLINE_ALIGNMENT
5178 Alignment required for trampolines, in bits.
5180 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
5181 is used for aligning trampolines.
5184 @hook TARGET_TRAMPOLINE_INIT
5185 This hook is called to initialize a trampoline.
5186 @var{m_tramp} is an RTX for the memory block for the trampoline; @var{fndecl}
5187 is the @code{FUNCTION_DECL} for the nested function; @var{static_chain} is an
5188 RTX for the static chain value that should be passed to the function
5191 If the target defines @code{TARGET_ASM_TRAMPOLINE_TEMPLATE}, then the
5192 first thing this hook should do is emit a block move into @var{m_tramp}
5193 from the memory block returned by @code{assemble_trampoline_template}.
5194 Note that the block move need only cover the constant parts of the
5195 trampoline. If the target isolates the variable parts of the trampoline
5196 to the end, not all @code{TRAMPOLINE_SIZE} bytes need be copied.
5198 If the target requires any other actions, such as flushing caches or
5199 enabling stack execution, these actions should be performed after
5200 initializing the trampoline proper.
5203 @hook TARGET_TRAMPOLINE_ADJUST_ADDRESS
5204 This hook should perform any machine-specific adjustment in
5205 the address of the trampoline. Its argument contains the address of the
5206 memory block that was passed to @code{TARGET_TRAMPOLINE_INIT}. In case
5207 the address to be used for a function call should be different from the
5208 address at which the template was stored, the different address should
5209 be returned; otherwise @var{addr} should be returned unchanged.
5210 If this hook is not defined, @var{addr} will be used for function calls.
5213 Implementing trampolines is difficult on many machines because they have
5214 separate instruction and data caches. Writing into a stack location
5215 fails to clear the memory in the instruction cache, so when the program
5216 jumps to that location, it executes the old contents.
5218 Here are two possible solutions. One is to clear the relevant parts of
5219 the instruction cache whenever a trampoline is set up. The other is to
5220 make all trampolines identical, by having them jump to a standard
5221 subroutine. The former technique makes trampoline execution faster; the
5222 latter makes initialization faster.
5224 To clear the instruction cache when a trampoline is initialized, define
5225 the following macro.
5227 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5228 If defined, expands to a C expression clearing the @emph{instruction
5229 cache} in the specified interval. The definition of this macro would
5230 typically be a series of @code{asm} statements. Both @var{beg} and
5231 @var{end} are both pointer expressions.
5234 The operating system may also require the stack to be made executable
5235 before calling the trampoline. To implement this requirement, define
5236 the following macro.
5238 @defmac ENABLE_EXECUTE_STACK
5239 Define this macro if certain operations must be performed before executing
5240 code located on the stack. The macro should expand to a series of C
5241 file-scope constructs (e.g.@: functions) and provide a unique entry point
5242 named @code{__enable_execute_stack}. The target is responsible for
5243 emitting calls to the entry point in the code, for example from the
5244 @code{TARGET_TRAMPOLINE_INIT} hook.
5247 To use a standard subroutine, define the following macro. In addition,
5248 you must make sure that the instructions in a trampoline fill an entire
5249 cache line with identical instructions, or else ensure that the
5250 beginning of the trampoline code is always aligned at the same point in
5251 its cache line. Look in @file{m68k.h} as a guide.
5253 @defmac TRANSFER_FROM_TRAMPOLINE
5254 Define this macro if trampolines need a special subroutine to do their
5255 work. The macro should expand to a series of @code{asm} statements
5256 which will be compiled with GCC@. They go in a library function named
5257 @code{__transfer_from_trampoline}.
5259 If you need to avoid executing the ordinary prologue code of a compiled
5260 C function when you jump to the subroutine, you can do so by placing a
5261 special label of your own in the assembler code. Use one @code{asm}
5262 statement to generate an assembler label, and another to make the label
5263 global. Then trampolines can use that label to jump directly to your
5264 special assembler code.
5268 @section Implicit Calls to Library Routines
5269 @cindex library subroutine names
5270 @cindex @file{libgcc.a}
5272 @c prevent bad page break with this line
5273 Here is an explanation of implicit calls to library routines.
5275 @defmac DECLARE_LIBRARY_RENAMES
5276 This macro, if defined, should expand to a piece of C code that will get
5277 expanded when compiling functions for libgcc.a. It can be used to
5278 provide alternate names for GCC's internal library functions if there
5279 are ABI-mandated names that the compiler should provide.
5282 @findex set_optab_libfunc
5283 @findex init_one_libfunc
5284 @hook TARGET_INIT_LIBFUNCS
5285 This hook should declare additional library routines or rename
5286 existing ones, using the functions @code{set_optab_libfunc} and
5287 @code{init_one_libfunc} defined in @file{optabs.c}.
5288 @code{init_optabs} calls this macro after initializing all the normal
5291 The default is to do nothing. Most ports don't need to define this hook.
5294 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5295 This macro should return @code{true} if the library routine that
5296 implements the floating point comparison operator @var{comparison} in
5297 mode @var{mode} will return a boolean, and @var{false} if it will
5300 GCC's own floating point libraries return tristates from the
5301 comparison operators, so the default returns false always. Most ports
5302 don't need to define this macro.
5305 @defmac TARGET_LIB_INT_CMP_BIASED
5306 This macro should evaluate to @code{true} if the integer comparison
5307 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5308 operand is smaller than the second, 1 to indicate that they are equal,
5309 and 2 to indicate that the first operand is greater than the second.
5310 If this macro evaluates to @code{false} the comparison functions return
5311 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5312 in @file{libgcc.a}, you do not need to define this macro.
5315 @cindex US Software GOFAST, floating point emulation library
5316 @cindex floating point emulation library, US Software GOFAST
5317 @cindex GOFAST, floating point emulation library
5318 @findex gofast_maybe_init_libfuncs
5319 @defmac US_SOFTWARE_GOFAST
5320 Define this macro if your system C library uses the US Software GOFAST
5321 library to provide floating point emulation.
5323 In addition to defining this macro, your architecture must set
5324 @code{TARGET_INIT_LIBFUNCS} to @code{gofast_maybe_init_libfuncs}, or
5325 else call that function from its version of that hook. It is defined
5326 in @file{config/gofast.h}, which must be included by your
5327 architecture's @file{@var{cpu}.c} file. See @file{sparc/sparc.c} for
5330 If this macro is defined, the
5331 @code{TARGET_FLOAT_LIB_COMPARE_RETURNS_BOOL} target hook must return
5332 false for @code{SFmode} and @code{DFmode} comparisons.
5335 @cindex @code{EDOM}, implicit usage
5338 The value of @code{EDOM} on the target machine, as a C integer constant
5339 expression. If you don't define this macro, GCC does not attempt to
5340 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5341 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5344 If you do not define @code{TARGET_EDOM}, then compiled code reports
5345 domain errors by calling the library function and letting it report the
5346 error. If mathematical functions on your system use @code{matherr} when
5347 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5348 that @code{matherr} is used normally.
5351 @cindex @code{errno}, implicit usage
5352 @defmac GEN_ERRNO_RTX
5353 Define this macro as a C expression to create an rtl expression that
5354 refers to the global ``variable'' @code{errno}. (On certain systems,
5355 @code{errno} may not actually be a variable.) If you don't define this
5356 macro, a reasonable default is used.
5359 @cindex C99 math functions, implicit usage
5360 @defmac TARGET_C99_FUNCTIONS
5361 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5362 @code{sinf} and similarly for other functions defined by C99 standard. The
5363 default is zero because a number of existing systems lack support for these
5364 functions in their runtime so this macro needs to be redefined to one on
5365 systems that do support the C99 runtime.
5368 @cindex sincos math function, implicit usage
5369 @defmac TARGET_HAS_SINCOS
5370 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5371 and @code{cos} with the same argument to a call to @code{sincos}. The
5372 default is zero. The target has to provide the following functions:
5374 void sincos(double x, double *sin, double *cos);
5375 void sincosf(float x, float *sin, float *cos);
5376 void sincosl(long double x, long double *sin, long double *cos);
5380 @defmac NEXT_OBJC_RUNTIME
5381 Define this macro to generate code for Objective-C message sending using
5382 the calling convention of the NeXT system. This calling convention
5383 involves passing the object, the selector and the method arguments all
5384 at once to the method-lookup library function.
5386 The default calling convention passes just the object and the selector
5387 to the lookup function, which returns a pointer to the method.
5390 @node Addressing Modes
5391 @section Addressing Modes
5392 @cindex addressing modes
5394 @c prevent bad page break with this line
5395 This is about addressing modes.
5397 @defmac HAVE_PRE_INCREMENT
5398 @defmacx HAVE_PRE_DECREMENT
5399 @defmacx HAVE_POST_INCREMENT
5400 @defmacx HAVE_POST_DECREMENT
5401 A C expression that is nonzero if the machine supports pre-increment,
5402 pre-decrement, post-increment, or post-decrement addressing respectively.
5405 @defmac HAVE_PRE_MODIFY_DISP
5406 @defmacx HAVE_POST_MODIFY_DISP
5407 A C expression that is nonzero if the machine supports pre- or
5408 post-address side-effect generation involving constants other than
5409 the size of the memory operand.
5412 @defmac HAVE_PRE_MODIFY_REG
5413 @defmacx HAVE_POST_MODIFY_REG
5414 A C expression that is nonzero if the machine supports pre- or
5415 post-address side-effect generation involving a register displacement.
5418 @defmac CONSTANT_ADDRESS_P (@var{x})
5419 A C expression that is 1 if the RTX @var{x} is a constant which
5420 is a valid address. On most machines the default definition of
5421 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
5422 is acceptable, but a few machines are more restrictive as to which
5423 constant addresses are supported.
5426 @defmac CONSTANT_P (@var{x})
5427 @code{CONSTANT_P}, which is defined by target-independent code,
5428 accepts integer-values expressions whose values are not explicitly
5429 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5430 expressions and @code{const} arithmetic expressions, in addition to
5431 @code{const_int} and @code{const_double} expressions.
5434 @defmac MAX_REGS_PER_ADDRESS
5435 A number, the maximum number of registers that can appear in a valid
5436 memory address. Note that it is up to you to specify a value equal to
5437 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5441 @hook TARGET_LEGITIMATE_ADDRESS_P
5442 A function that returns whether @var{x} (an RTX) is a legitimate memory
5443 address on the target machine for a memory operand of mode @var{mode}.
5445 Legitimate addresses are defined in two variants: a strict variant and a
5446 non-strict one. The @var{strict} parameter chooses which variant is
5447 desired by the caller.
5449 The strict variant is used in the reload pass. It must be defined so
5450 that any pseudo-register that has not been allocated a hard register is
5451 considered a memory reference. This is because in contexts where some
5452 kind of register is required, a pseudo-register with no hard register
5453 must be rejected. For non-hard registers, the strict variant should look
5454 up the @code{reg_renumber} array; it should then proceed using the hard
5455 register number in the array, or treat the pseudo as a memory reference
5456 if the array holds @code{-1}.
5458 The non-strict variant is used in other passes. It must be defined to
5459 accept all pseudo-registers in every context where some kind of
5460 register is required.
5462 Normally, constant addresses which are the sum of a @code{symbol_ref}
5463 and an integer are stored inside a @code{const} RTX to mark them as
5464 constant. Therefore, there is no need to recognize such sums
5465 specifically as legitimate addresses. Normally you would simply
5466 recognize any @code{const} as legitimate.
5468 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5469 sums that are not marked with @code{const}. It assumes that a naked
5470 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5471 naked constant sums as illegitimate addresses, so that none of them will
5472 be given to @code{PRINT_OPERAND_ADDRESS}.
5474 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5475 On some machines, whether a symbolic address is legitimate depends on
5476 the section that the address refers to. On these machines, define the
5477 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5478 into the @code{symbol_ref}, and then check for it here. When you see a
5479 @code{const}, you will have to look inside it to find the
5480 @code{symbol_ref} in order to determine the section. @xref{Assembler
5483 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5484 Some ports are still using a deprecated legacy substitute for
5485 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5489 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5493 and should @code{goto @var{label}} if the address @var{x} is a valid
5494 address on the target machine for a memory operand of mode @var{mode}.
5495 Whether the strict or non-strict variants are desired is defined by
5496 the @code{REG_OK_STRICT} macro introduced earlier in this section.
5497 Using the hook is usually simpler because it limits the number of
5498 files that are recompiled when changes are made.
5501 @defmac TARGET_MEM_CONSTRAINT
5502 A single character to be used instead of the default @code{'m'}
5503 character for general memory addresses. This defines the constraint
5504 letter which matches the memory addresses accepted by
5505 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5506 support new address formats in your back end without changing the
5507 semantics of the @code{'m'} constraint. This is necessary in order to
5508 preserve functionality of inline assembly constructs using the
5509 @code{'m'} constraint.
5512 @defmac FIND_BASE_TERM (@var{x})
5513 A C expression to determine the base term of address @var{x},
5514 or to provide a simplified version of @var{x} from which @file{alias.c}
5515 can easily find the base term. This macro is used in only two places:
5516 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5518 It is always safe for this macro to not be defined. It exists so
5519 that alias analysis can understand machine-dependent addresses.
5521 The typical use of this macro is to handle addresses containing
5522 a label_ref or symbol_ref within an UNSPEC@.
5525 @hook TARGET_LEGITIMIZE_ADDRESS
5526 This hook is given an invalid memory address @var{x} for an
5527 operand of mode @var{mode} and should try to return a valid memory
5530 @findex break_out_memory_refs
5531 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5532 and @var{oldx} will be the operand that was given to that function to produce
5535 The code of the hook should not alter the substructure of
5536 @var{x}. If it transforms @var{x} into a more legitimate form, it
5537 should return the new @var{x}.
5539 It is not necessary for this hook to come up with a legitimate address.
5540 The compiler has standard ways of doing so in all cases. In fact, it
5541 is safe to omit this hook or make it return @var{x} if it cannot find
5542 a valid way to legitimize the address. But often a machine-dependent
5543 strategy can generate better code.
5546 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5547 A C compound statement that attempts to replace @var{x}, which is an address
5548 that needs reloading, with a valid memory address for an operand of mode
5549 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5550 It is not necessary to define this macro, but it might be useful for
5551 performance reasons.
5553 For example, on the i386, it is sometimes possible to use a single
5554 reload register instead of two by reloading a sum of two pseudo
5555 registers into a register. On the other hand, for number of RISC
5556 processors offsets are limited so that often an intermediate address
5557 needs to be generated in order to address a stack slot. By defining
5558 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5559 generated for adjacent some stack slots can be made identical, and thus
5562 @emph{Note}: This macro should be used with caution. It is necessary
5563 to know something of how reload works in order to effectively use this,
5564 and it is quite easy to produce macros that build in too much knowledge
5565 of reload internals.
5567 @emph{Note}: This macro must be able to reload an address created by a
5568 previous invocation of this macro. If it fails to handle such addresses
5569 then the compiler may generate incorrect code or abort.
5572 The macro definition should use @code{push_reload} to indicate parts that
5573 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5574 suitable to be passed unaltered to @code{push_reload}.
5576 The code generated by this macro must not alter the substructure of
5577 @var{x}. If it transforms @var{x} into a more legitimate form, it
5578 should assign @var{x} (which will always be a C variable) a new value.
5579 This also applies to parts that you change indirectly by calling
5582 @findex strict_memory_address_p
5583 The macro definition may use @code{strict_memory_address_p} to test if
5584 the address has become legitimate.
5587 If you want to change only a part of @var{x}, one standard way of doing
5588 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5589 single level of rtl. Thus, if the part to be changed is not at the
5590 top level, you'll need to replace first the top level.
5591 It is not necessary for this macro to come up with a legitimate
5592 address; but often a machine-dependent strategy can generate better code.
5595 @hook TARGET_MODE_DEPENDENT_ADDRESS_P
5596 This hook returns @code{true} if memory address @var{addr} can have
5597 different meanings depending on the machine mode of the memory
5598 reference it is used for or if the address is valid for some modes
5601 Autoincrement and autodecrement addresses typically have mode-dependent
5602 effects because the amount of the increment or decrement is the size
5603 of the operand being addressed. Some machines have other mode-dependent
5604 addresses. Many RISC machines have no mode-dependent addresses.
5606 You may assume that @var{addr} is a valid address for the machine.
5608 The default version of this hook returns @code{false}.
5611 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5612 A C statement or compound statement with a conditional @code{goto
5613 @var{label};} executed if memory address @var{x} (an RTX) can have
5614 different meanings depending on the machine mode of the memory
5615 reference it is used for or if the address is valid for some modes
5618 Autoincrement and autodecrement addresses typically have mode-dependent
5619 effects because the amount of the increment or decrement is the size
5620 of the operand being addressed. Some machines have other mode-dependent
5621 addresses. Many RISC machines have no mode-dependent addresses.
5623 You may assume that @var{addr} is a valid address for the machine.
5625 These are obsolete macros, replaced by the
5626 @code{TARGET_MODE_DEPENDENT_ADDRESS_P} target hook.
5629 @defmac LEGITIMATE_CONSTANT_P (@var{x})
5630 A C expression that is nonzero if @var{x} is a legitimate constant for
5631 an immediate operand on the target machine. You can assume that
5632 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
5633 @samp{1} is a suitable definition for this macro on machines where
5634 anything @code{CONSTANT_P} is valid.
5637 @hook TARGET_DELEGITIMIZE_ADDRESS
5638 This hook is used to undo the possibly obfuscating effects of the
5639 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5640 macros. Some backend implementations of these macros wrap symbol
5641 references inside an @code{UNSPEC} rtx to represent PIC or similar
5642 addressing modes. This target hook allows GCC's optimizers to understand
5643 the semantics of these opaque @code{UNSPEC}s by converting them back
5644 into their original form.
5647 @hook TARGET_CANNOT_FORCE_CONST_MEM
5648 This hook should return true if @var{x} is of a form that cannot (or
5649 should not) be spilled to the constant pool. The default version of
5650 this hook returns false.
5652 The primary reason to define this hook is to prevent reload from
5653 deciding that a non-legitimate constant would be better reloaded
5654 from the constant pool instead of spilling and reloading a register
5655 holding the constant. This restriction is often true of addresses
5656 of TLS symbols for various targets.
5659 @hook TARGET_USE_BLOCKS_FOR_CONSTANT_P
5660 This hook should return true if pool entries for constant @var{x} can
5661 be placed in an @code{object_block} structure. @var{mode} is the mode
5664 The default version returns false for all constants.
5667 @hook TARGET_BUILTIN_RECIPROCAL
5668 This hook should return the DECL of a function that implements reciprocal of
5669 the builtin function with builtin function code @var{fn}, or
5670 @code{NULL_TREE} if such a function is not available. @var{md_fn} is true
5671 when @var{fn} is a code of a machine-dependent builtin function. When
5672 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5673 of a square root function are performed, and only reciprocals of @code{sqrt}
5677 @hook TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD
5678 This hook should return the DECL of a function @var{f} that given an
5679 address @var{addr} as an argument returns a mask @var{m} that can be
5680 used to extract from two vectors the relevant data that resides in
5681 @var{addr} in case @var{addr} is not properly aligned.
5683 The autovectorizer, when vectorizing a load operation from an address
5684 @var{addr} that may be unaligned, will generate two vector loads from
5685 the two aligned addresses around @var{addr}. It then generates a
5686 @code{REALIGN_LOAD} operation to extract the relevant data from the
5687 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5688 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5689 the third argument, @var{OFF}, defines how the data will be extracted
5690 from these two vectors: if @var{OFF} is 0, then the returned vector is
5691 @var{v2}; otherwise, the returned vector is composed from the last
5692 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5693 @var{OFF} elements of @var{v2}.
5695 If this hook is defined, the autovectorizer will generate a call
5696 to @var{f} (using the DECL tree that this hook returns) and will
5697 use the return value of @var{f} as the argument @var{OFF} to
5698 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5699 should comply with the semantics expected by @code{REALIGN_LOAD}
5701 If this hook is not defined, then @var{addr} will be used as
5702 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5703 log2(@var{VS}) @minus{} 1 bits of @var{addr} will be considered.
5706 @hook TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN
5707 This hook should return the DECL of a function @var{f} that implements
5708 widening multiplication of the even elements of two input vectors of type @var{x}.
5710 If this hook is defined, the autovectorizer will use it along with the
5711 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD} target hook when vectorizing
5712 widening multiplication in cases that the order of the results does not have to be
5713 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5714 @code{widen_mult_hi/lo} idioms will be used.
5717 @hook TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD
5718 This hook should return the DECL of a function @var{f} that implements
5719 widening multiplication of the odd elements of two input vectors of type @var{x}.
5721 If this hook is defined, the autovectorizer will use it along with the
5722 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN} target hook when vectorizing
5723 widening multiplication in cases that the order of the results does not have to be
5724 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5725 @code{widen_mult_hi/lo} idioms will be used.
5728 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST
5729 Returns cost of different scalar or vector statements for vectorization cost model.
5730 For vector memory operations the cost may depend on type (@var{vectype}) and
5731 misalignment value (@var{misalign}).
5734 @hook TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE
5735 Return true if vector alignment is reachable (by peeling N iterations) for the given type.
5738 @hook TARGET_VECTORIZE_BUILTIN_VEC_PERM
5739 Target builtin that implements vector permute.
5742 @hook TARGET_VECTORIZE_BUILTIN_VEC_PERM_OK
5743 Return true if a vector created for @code{builtin_vec_perm} is valid.
5746 @hook TARGET_VECTORIZE_BUILTIN_CONVERSION
5747 This hook should return the DECL of a function that implements conversion of the
5748 input vector of type @var{src_type} to type @var{dest_type}.
5749 The value of @var{code} is one of the enumerators in @code{enum tree_code} and
5750 specifies how the conversion is to be applied
5751 (truncation, rounding, etc.).
5753 If this hook is defined, the autovectorizer will use the
5754 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5755 conversion. Otherwise, it will return @code{NULL_TREE}.
5758 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION
5759 This hook should return the decl of a function that implements the
5760 vectorized variant of the builtin function with builtin function code
5761 @var{code} or @code{NULL_TREE} if such a function is not available.
5762 The value of @var{fndecl} is the builtin function declaration. The
5763 return type of the vectorized function shall be of vector type
5764 @var{vec_type_out} and the argument types should be @var{vec_type_in}.
5767 @hook TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT
5768 This hook should return true if the target supports misaligned vector
5769 store/load of a specific factor denoted in the @var{misalignment}
5770 parameter. The vector store/load should be of machine mode @var{mode} and
5771 the elements in the vectors should be of type @var{type}. @var{is_packed}
5772 parameter is true if the memory access is defined in a packed struct.
5775 @node Anchored Addresses
5776 @section Anchored Addresses
5777 @cindex anchored addresses
5778 @cindex @option{-fsection-anchors}
5780 GCC usually addresses every static object as a separate entity.
5781 For example, if we have:
5785 int foo (void) @{ return a + b + c; @}
5788 the code for @code{foo} will usually calculate three separate symbolic
5789 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5790 it would be better to calculate just one symbolic address and access
5791 the three variables relative to it. The equivalent pseudocode would
5797 register int *xr = &x;
5798 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5802 (which isn't valid C). We refer to shared addresses like @code{x} as
5803 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5805 The hooks below describe the target properties that GCC needs to know
5806 in order to make effective use of section anchors. It won't use
5807 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5808 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5810 @hook TARGET_MIN_ANCHOR_OFFSET
5811 The minimum offset that should be applied to a section anchor.
5812 On most targets, it should be the smallest offset that can be
5813 applied to a base register while still giving a legitimate address
5814 for every mode. The default value is 0.
5817 @hook TARGET_MAX_ANCHOR_OFFSET
5818 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5819 offset that should be applied to section anchors. The default
5823 @hook TARGET_ASM_OUTPUT_ANCHOR
5824 Write the assembly code to define section anchor @var{x}, which is a
5825 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5826 The hook is called with the assembly output position set to the beginning
5827 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5829 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5830 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5831 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5832 is @code{NULL}, which disables the use of section anchors altogether.
5835 @hook TARGET_USE_ANCHORS_FOR_SYMBOL_P
5836 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5837 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5838 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5840 The default version is correct for most targets, but you might need to
5841 intercept this hook to handle things like target-specific attributes
5842 or target-specific sections.
5845 @node Condition Code
5846 @section Condition Code Status
5847 @cindex condition code status
5849 The macros in this section can be split in two families, according to the
5850 two ways of representing condition codes in GCC.
5852 The first representation is the so called @code{(cc0)} representation
5853 (@pxref{Jump Patterns}), where all instructions can have an implicit
5854 clobber of the condition codes. The second is the condition code
5855 register representation, which provides better schedulability for
5856 architectures that do have a condition code register, but on which
5857 most instructions do not affect it. The latter category includes
5860 The implicit clobbering poses a strong restriction on the placement of
5861 the definition and use of the condition code, which need to be in adjacent
5862 insns for machines using @code{(cc0)}. This can prevent important
5863 optimizations on some machines. For example, on the IBM RS/6000, there
5864 is a delay for taken branches unless the condition code register is set
5865 three instructions earlier than the conditional branch. The instruction
5866 scheduler cannot perform this optimization if it is not permitted to
5867 separate the definition and use of the condition code register.
5869 For this reason, it is possible and suggested to use a register to
5870 represent the condition code for new ports. If there is a specific
5871 condition code register in the machine, use a hard register. If the
5872 condition code or comparison result can be placed in any general register,
5873 or if there are multiple condition registers, use a pseudo register.
5874 Registers used to store the condition code value will usually have a mode
5875 that is in class @code{MODE_CC}.
5877 Alternatively, you can use @code{BImode} if the comparison operator is
5878 specified already in the compare instruction. In this case, you are not
5879 interested in most macros in this section.
5882 * CC0 Condition Codes:: Old style representation of condition codes.
5883 * MODE_CC Condition Codes:: Modern representation of condition codes.
5884 * Cond. Exec. Macros:: Macros to control conditional execution.
5887 @node CC0 Condition Codes
5888 @subsection Representation of condition codes using @code{(cc0)}
5892 The file @file{conditions.h} defines a variable @code{cc_status} to
5893 describe how the condition code was computed (in case the interpretation of
5894 the condition code depends on the instruction that it was set by). This
5895 variable contains the RTL expressions on which the condition code is
5896 currently based, and several standard flags.
5898 Sometimes additional machine-specific flags must be defined in the machine
5899 description header file. It can also add additional machine-specific
5900 information by defining @code{CC_STATUS_MDEP}.
5902 @defmac CC_STATUS_MDEP
5903 C code for a data type which is used for declaring the @code{mdep}
5904 component of @code{cc_status}. It defaults to @code{int}.
5906 This macro is not used on machines that do not use @code{cc0}.
5909 @defmac CC_STATUS_MDEP_INIT
5910 A C expression to initialize the @code{mdep} field to ``empty''.
5911 The default definition does nothing, since most machines don't use
5912 the field anyway. If you want to use the field, you should probably
5913 define this macro to initialize it.
5915 This macro is not used on machines that do not use @code{cc0}.
5918 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5919 A C compound statement to set the components of @code{cc_status}
5920 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5921 this macro's responsibility to recognize insns that set the condition
5922 code as a byproduct of other activity as well as those that explicitly
5925 This macro is not used on machines that do not use @code{cc0}.
5927 If there are insns that do not set the condition code but do alter
5928 other machine registers, this macro must check to see whether they
5929 invalidate the expressions that the condition code is recorded as
5930 reflecting. For example, on the 68000, insns that store in address
5931 registers do not set the condition code, which means that usually
5932 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5933 insns. But suppose that the previous insn set the condition code
5934 based on location @samp{a4@@(102)} and the current insn stores a new
5935 value in @samp{a4}. Although the condition code is not changed by
5936 this, it will no longer be true that it reflects the contents of
5937 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5938 @code{cc_status} in this case to say that nothing is known about the
5939 condition code value.
5941 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5942 with the results of peephole optimization: insns whose patterns are
5943 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5944 constants which are just the operands. The RTL structure of these
5945 insns is not sufficient to indicate what the insns actually do. What
5946 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5947 @code{CC_STATUS_INIT}.
5949 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5950 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5951 @samp{cc}. This avoids having detailed information about patterns in
5952 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5955 @node MODE_CC Condition Codes
5956 @subsection Representation of condition codes using registers
5960 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5961 On many machines, the condition code may be produced by other instructions
5962 than compares, for example the branch can use directly the condition
5963 code set by a subtract instruction. However, on some machines
5964 when the condition code is set this way some bits (such as the overflow
5965 bit) are not set in the same way as a test instruction, so that a different
5966 branch instruction must be used for some conditional branches. When
5967 this happens, use the machine mode of the condition code register to
5968 record different formats of the condition code register. Modes can
5969 also be used to record which compare instruction (e.g. a signed or an
5970 unsigned comparison) produced the condition codes.
5972 If other modes than @code{CCmode} are required, add them to
5973 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5974 a mode given an operand of a compare. This is needed because the modes
5975 have to be chosen not only during RTL generation but also, for example,
5976 by instruction combination. The result of @code{SELECT_CC_MODE} should
5977 be consistent with the mode used in the patterns; for example to support
5978 the case of the add on the SPARC discussed above, we have the pattern
5982 [(set (reg:CC_NOOV 0)
5984 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5985 (match_operand:SI 1 "arith_operand" "rI"))
5992 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5993 for comparisons whose argument is a @code{plus}:
5996 #define SELECT_CC_MODE(OP,X,Y) \
5997 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5998 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5999 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
6000 || GET_CODE (X) == NEG) \
6001 ? CC_NOOVmode : CCmode))
6004 Another reason to use modes is to retain information on which operands
6005 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
6008 You should define this macro if and only if you define extra CC modes
6009 in @file{@var{machine}-modes.def}.
6012 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
6013 On some machines not all possible comparisons are defined, but you can
6014 convert an invalid comparison into a valid one. For example, the Alpha
6015 does not have a @code{GT} comparison, but you can use an @code{LT}
6016 comparison instead and swap the order of the operands.
6018 On such machines, define this macro to be a C statement to do any
6019 required conversions. @var{code} is the initial comparison code
6020 and @var{op0} and @var{op1} are the left and right operands of the
6021 comparison, respectively. You should modify @var{code}, @var{op0}, and
6022 @var{op1} as required.
6024 GCC will not assume that the comparison resulting from this macro is
6025 valid but will see if the resulting insn matches a pattern in the
6028 You need not define this macro if it would never change the comparison
6032 @defmac REVERSIBLE_CC_MODE (@var{mode})
6033 A C expression whose value is one if it is always safe to reverse a
6034 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
6035 can ever return @var{mode} for a floating-point inequality comparison,
6036 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
6038 You need not define this macro if it would always returns zero or if the
6039 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
6040 For example, here is the definition used on the SPARC, where floating-point
6041 inequality comparisons are always given @code{CCFPEmode}:
6044 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
6048 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
6049 A C expression whose value is reversed condition code of the @var{code} for
6050 comparison done in CC_MODE @var{mode}. The macro is used only in case
6051 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
6052 machine has some non-standard way how to reverse certain conditionals. For
6053 instance in case all floating point conditions are non-trapping, compiler may
6054 freely convert unordered compares to ordered one. Then definition may look
6058 #define REVERSE_CONDITION(CODE, MODE) \
6059 ((MODE) != CCFPmode ? reverse_condition (CODE) \
6060 : reverse_condition_maybe_unordered (CODE))
6064 @hook TARGET_FIXED_CONDITION_CODE_REGS
6065 On targets which do not use @code{(cc0)}, and which use a hard
6066 register rather than a pseudo-register to hold condition codes, the
6067 regular CSE passes are often not able to identify cases in which the
6068 hard register is set to a common value. Use this hook to enable a
6069 small pass which optimizes such cases. This hook should return true
6070 to enable this pass, and it should set the integers to which its
6071 arguments point to the hard register numbers used for condition codes.
6072 When there is only one such register, as is true on most systems, the
6073 integer pointed to by @var{p2} should be set to
6074 @code{INVALID_REGNUM}.
6076 The default version of this hook returns false.
6079 @hook TARGET_CC_MODES_COMPATIBLE
6080 On targets which use multiple condition code modes in class
6081 @code{MODE_CC}, it is sometimes the case that a comparison can be
6082 validly done in more than one mode. On such a system, define this
6083 target hook to take two mode arguments and to return a mode in which
6084 both comparisons may be validly done. If there is no such mode,
6085 return @code{VOIDmode}.
6087 The default version of this hook checks whether the modes are the
6088 same. If they are, it returns that mode. If they are different, it
6089 returns @code{VOIDmode}.
6092 @node Cond. Exec. Macros
6093 @subsection Macros to control conditional execution
6094 @findex conditional execution
6097 There is one macro that may need to be defined for targets
6098 supporting conditional execution, independent of how they
6099 represent conditional branches.
6101 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2})
6102 A C expression that returns true if the conditional execution predicate
6103 @var{op1}, a comparison operation, is the inverse of @var{op2} and vice
6104 versa. Define this to return 0 if the target has conditional execution
6105 predicates that cannot be reversed safely. There is no need to validate
6106 that the arguments of op1 and op2 are the same, this is done separately.
6107 If no expansion is specified, this macro is defined as follows:
6110 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
6111 (GET_CODE ((x)) == reversed_comparison_code ((y), NULL))
6116 @section Describing Relative Costs of Operations
6117 @cindex costs of instructions
6118 @cindex relative costs
6119 @cindex speed of instructions
6121 These macros let you describe the relative speed of various operations
6122 on the target machine.
6124 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6125 A C expression for the cost of moving data of mode @var{mode} from a
6126 register in class @var{from} to one in class @var{to}. The classes are
6127 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6128 value of 2 is the default; other values are interpreted relative to
6131 It is not required that the cost always equal 2 when @var{from} is the
6132 same as @var{to}; on some machines it is expensive to move between
6133 registers if they are not general registers.
6135 If reload sees an insn consisting of a single @code{set} between two
6136 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6137 classes returns a value of 2, reload does not check to ensure that the
6138 constraints of the insn are met. Setting a cost of other than 2 will
6139 allow reload to verify that the constraints are met. You should do this
6140 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6142 These macros are obsolete, new ports should use the target hook
6143 @code{TARGET_REGISTER_MOVE_COST} instead.
6146 @hook TARGET_REGISTER_MOVE_COST
6147 This target hook should return the cost of moving data of mode @var{mode}
6148 from a register in class @var{from} to one in class @var{to}. The classes
6149 are expressed using the enumeration values such as @code{GENERAL_REGS}.
6150 A value of 2 is the default; other values are interpreted relative to
6153 It is not required that the cost always equal 2 when @var{from} is the
6154 same as @var{to}; on some machines it is expensive to move between
6155 registers if they are not general registers.
6157 If reload sees an insn consisting of a single @code{set} between two
6158 hard registers, and if @code{TARGET_REGISTER_MOVE_COST} applied to their
6159 classes returns a value of 2, reload does not check to ensure that the
6160 constraints of the insn are met. Setting a cost of other than 2 will
6161 allow reload to verify that the constraints are met. You should do this
6162 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6164 The default version of this function returns 2.
6167 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6168 A C expression for the cost of moving data of mode @var{mode} between a
6169 register of class @var{class} and memory; @var{in} is zero if the value
6170 is to be written to memory, nonzero if it is to be read in. This cost
6171 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6172 registers and memory is more expensive than between two registers, you
6173 should define this macro to express the relative cost.
6175 If you do not define this macro, GCC uses a default cost of 4 plus
6176 the cost of copying via a secondary reload register, if one is
6177 needed. If your machine requires a secondary reload register to copy
6178 between memory and a register of @var{class} but the reload mechanism is
6179 more complex than copying via an intermediate, define this macro to
6180 reflect the actual cost of the move.
6182 GCC defines the function @code{memory_move_secondary_cost} if
6183 secondary reloads are needed. It computes the costs due to copying via
6184 a secondary register. If your machine copies from memory using a
6185 secondary register in the conventional way but the default base value of
6186 4 is not correct for your machine, define this macro to add some other
6187 value to the result of that function. The arguments to that function
6188 are the same as to this macro.
6190 These macros are obsolete, new ports should use the target hook
6191 @code{TARGET_MEMORY_MOVE_COST} instead.
6194 @hook TARGET_MEMORY_MOVE_COST
6195 This target hook should return the cost of moving data of mode @var{mode}
6196 between a register of class @var{rclass} and memory; @var{in} is @code{false}
6197 if the value is to be written to memory, @code{true} if it is to be read in.
6198 This cost is relative to those in @code{TARGET_REGISTER_MOVE_COST}.
6199 If moving between registers and memory is more expensive than between two
6200 registers, you should add this target hook to express the relative cost.
6202 If you do not add this target hook, GCC uses a default cost of 4 plus
6203 the cost of copying via a secondary reload register, if one is
6204 needed. If your machine requires a secondary reload register to copy
6205 between memory and a register of @var{rclass} but the reload mechanism is
6206 more complex than copying via an intermediate, use this target hook to
6207 reflect the actual cost of the move.
6209 GCC defines the function @code{memory_move_secondary_cost} if
6210 secondary reloads are needed. It computes the costs due to copying via
6211 a secondary register. If your machine copies from memory using a
6212 secondary register in the conventional way but the default base value of
6213 4 is not correct for your machine, use this target hook to add some other
6214 value to the result of that function. The arguments to that function
6215 are the same as to this target hook.
6218 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6219 A C expression for the cost of a branch instruction. A value of 1 is the
6220 default; other values are interpreted relative to that. Parameter @var{speed_p}
6221 is true when the branch in question should be optimized for speed. When
6222 it is false, @code{BRANCH_COST} should be returning value optimal for code size
6223 rather then performance considerations. @var{predictable_p} is true for well
6224 predictable branches. On many architectures the @code{BRANCH_COST} can be
6228 Here are additional macros which do not specify precise relative costs,
6229 but only that certain actions are more expensive than GCC would
6232 @defmac SLOW_BYTE_ACCESS
6233 Define this macro as a C expression which is nonzero if accessing less
6234 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6235 faster than accessing a word of memory, i.e., if such access
6236 require more than one instruction or if there is no difference in cost
6237 between byte and (aligned) word loads.
6239 When this macro is not defined, the compiler will access a field by
6240 finding the smallest containing object; when it is defined, a fullword
6241 load will be used if alignment permits. Unless bytes accesses are
6242 faster than word accesses, using word accesses is preferable since it
6243 may eliminate subsequent memory access if subsequent accesses occur to
6244 other fields in the same word of the structure, but to different bytes.
6247 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6248 Define this macro to be the value 1 if memory accesses described by the
6249 @var{mode} and @var{alignment} parameters have a cost many times greater
6250 than aligned accesses, for example if they are emulated in a trap
6253 When this macro is nonzero, the compiler will act as if
6254 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6255 moves. This can cause significantly more instructions to be produced.
6256 Therefore, do not set this macro nonzero if unaligned accesses only add a
6257 cycle or two to the time for a memory access.
6259 If the value of this macro is always zero, it need not be defined. If
6260 this macro is defined, it should produce a nonzero value when
6261 @code{STRICT_ALIGNMENT} is nonzero.
6264 @defmac MOVE_RATIO (@var{speed})
6265 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6266 which a sequence of insns should be generated instead of a
6267 string move insn or a library call. Increasing the value will always
6268 make code faster, but eventually incurs high cost in increased code size.
6270 Note that on machines where the corresponding move insn is a
6271 @code{define_expand} that emits a sequence of insns, this macro counts
6272 the number of such sequences.
6274 The parameter @var{speed} is true if the code is currently being
6275 optimized for speed rather than size.
6277 If you don't define this, a reasonable default is used.
6280 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6281 A C expression used to determine whether @code{move_by_pieces} will be used to
6282 copy a chunk of memory, or whether some other block move mechanism
6283 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6284 than @code{MOVE_RATIO}.
6287 @defmac MOVE_MAX_PIECES
6288 A C expression used by @code{move_by_pieces} to determine the largest unit
6289 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6292 @defmac CLEAR_RATIO (@var{speed})
6293 The threshold of number of scalar move insns, @emph{below} which a sequence
6294 of insns should be generated to clear memory instead of a string clear insn
6295 or a library call. Increasing the value will always make code faster, but
6296 eventually incurs high cost in increased code size.
6298 The parameter @var{speed} is true if the code is currently being
6299 optimized for speed rather than size.
6301 If you don't define this, a reasonable default is used.
6304 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6305 A C expression used to determine whether @code{clear_by_pieces} will be used
6306 to clear a chunk of memory, or whether some other block clear mechanism
6307 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6308 than @code{CLEAR_RATIO}.
6311 @defmac SET_RATIO (@var{speed})
6312 The threshold of number of scalar move insns, @emph{below} which a sequence
6313 of insns should be generated to set memory to a constant value, instead of
6314 a block set insn or a library call.
6315 Increasing the value will always make code faster, but
6316 eventually incurs high cost in increased code size.
6318 The parameter @var{speed} is true if the code is currently being
6319 optimized for speed rather than size.
6321 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6324 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6325 A C expression used to determine whether @code{store_by_pieces} will be
6326 used to set a chunk of memory to a constant value, or whether some
6327 other mechanism will be used. Used by @code{__builtin_memset} when
6328 storing values other than constant zero.
6329 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6330 than @code{SET_RATIO}.
6333 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6334 A C expression used to determine whether @code{store_by_pieces} will be
6335 used to set a chunk of memory to a constant string value, or whether some
6336 other mechanism will be used. Used by @code{__builtin_strcpy} when
6337 called with a constant source string.
6338 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6339 than @code{MOVE_RATIO}.
6342 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6343 A C expression used to determine whether a load postincrement is a good
6344 thing to use for a given mode. Defaults to the value of
6345 @code{HAVE_POST_INCREMENT}.
6348 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6349 A C expression used to determine whether a load postdecrement is a good
6350 thing to use for a given mode. Defaults to the value of
6351 @code{HAVE_POST_DECREMENT}.
6354 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6355 A C expression used to determine whether a load preincrement is a good
6356 thing to use for a given mode. Defaults to the value of
6357 @code{HAVE_PRE_INCREMENT}.
6360 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6361 A C expression used to determine whether a load predecrement is a good
6362 thing to use for a given mode. Defaults to the value of
6363 @code{HAVE_PRE_DECREMENT}.
6366 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6367 A C expression used to determine whether a store postincrement is a good
6368 thing to use for a given mode. Defaults to the value of
6369 @code{HAVE_POST_INCREMENT}.
6372 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6373 A C expression used to determine whether a store postdecrement is a good
6374 thing to use for a given mode. Defaults to the value of
6375 @code{HAVE_POST_DECREMENT}.
6378 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6379 This macro is used to determine whether a store preincrement is a good
6380 thing to use for a given mode. Defaults to the value of
6381 @code{HAVE_PRE_INCREMENT}.
6384 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6385 This macro is used to determine whether a store predecrement is a good
6386 thing to use for a given mode. Defaults to the value of
6387 @code{HAVE_PRE_DECREMENT}.
6390 @defmac NO_FUNCTION_CSE
6391 Define this macro if it is as good or better to call a constant
6392 function address than to call an address kept in a register.
6395 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
6396 Define this macro if a non-short-circuit operation produced by
6397 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6398 @code{BRANCH_COST} is greater than or equal to the value 2.
6401 @hook TARGET_RTX_COSTS
6402 This target hook describes the relative costs of RTL expressions.
6404 The cost may depend on the precise form of the expression, which is
6405 available for examination in @var{x}, and the rtx code of the expression
6406 in which it is contained, found in @var{outer_code}. @var{code} is the
6407 expression code---redundant, since it can be obtained with
6408 @code{GET_CODE (@var{x})}.
6410 In implementing this hook, you can use the construct
6411 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6414 On entry to the hook, @code{*@var{total}} contains a default estimate
6415 for the cost of the expression. The hook should modify this value as
6416 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6417 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6418 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6420 When optimizing for code size, i.e.@: when @code{speed} is
6421 false, this target hook should be used to estimate the relative
6422 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6424 The hook returns true when all subexpressions of @var{x} have been
6425 processed, and false when @code{rtx_cost} should recurse.
6428 @hook TARGET_ADDRESS_COST
6429 This hook computes the cost of an addressing mode that contains
6430 @var{address}. If not defined, the cost is computed from
6431 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6433 For most CISC machines, the default cost is a good approximation of the
6434 true cost of the addressing mode. However, on RISC machines, all
6435 instructions normally have the same length and execution time. Hence
6436 all addresses will have equal costs.
6438 In cases where more than one form of an address is known, the form with
6439 the lowest cost will be used. If multiple forms have the same, lowest,
6440 cost, the one that is the most complex will be used.
6442 For example, suppose an address that is equal to the sum of a register
6443 and a constant is used twice in the same basic block. When this macro
6444 is not defined, the address will be computed in a register and memory
6445 references will be indirect through that register. On machines where
6446 the cost of the addressing mode containing the sum is no higher than
6447 that of a simple indirect reference, this will produce an additional
6448 instruction and possibly require an additional register. Proper
6449 specification of this macro eliminates this overhead for such machines.
6451 This hook is never called with an invalid address.
6453 On machines where an address involving more than one register is as
6454 cheap as an address computation involving only one register, defining
6455 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6456 be live over a region of code where only one would have been if
6457 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6458 should be considered in the definition of this macro. Equivalent costs
6459 should probably only be given to addresses with different numbers of
6460 registers on machines with lots of registers.
6464 @section Adjusting the Instruction Scheduler
6466 The instruction scheduler may need a fair amount of machine-specific
6467 adjustment in order to produce good code. GCC provides several target
6468 hooks for this purpose. It is usually enough to define just a few of
6469 them: try the first ones in this list first.
6471 @hook TARGET_SCHED_ISSUE_RATE
6472 This hook returns the maximum number of instructions that can ever
6473 issue at the same time on the target machine. The default is one.
6474 Although the insn scheduler can define itself the possibility of issue
6475 an insn on the same cycle, the value can serve as an additional
6476 constraint to issue insns on the same simulated processor cycle (see
6477 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6478 This value must be constant over the entire compilation. If you need
6479 it to vary depending on what the instructions are, you must use
6480 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6483 @hook TARGET_SCHED_VARIABLE_ISSUE
6484 This hook is executed by the scheduler after it has scheduled an insn
6485 from the ready list. It should return the number of insns which can
6486 still be issued in the current cycle. The default is
6487 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6488 @code{USE}, which normally are not counted against the issue rate.
6489 You should define this hook if some insns take more machine resources
6490 than others, so that fewer insns can follow them in the same cycle.
6491 @var{file} is either a null pointer, or a stdio stream to write any
6492 debug output to. @var{verbose} is the verbose level provided by
6493 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6497 @hook TARGET_SCHED_ADJUST_COST
6498 This function corrects the value of @var{cost} based on the
6499 relationship between @var{insn} and @var{dep_insn} through the
6500 dependence @var{link}. It should return the new value. The default
6501 is to make no adjustment to @var{cost}. This can be used for example
6502 to specify to the scheduler using the traditional pipeline description
6503 that an output- or anti-dependence does not incur the same cost as a
6504 data-dependence. If the scheduler using the automaton based pipeline
6505 description, the cost of anti-dependence is zero and the cost of
6506 output-dependence is maximum of one and the difference of latency
6507 times of the first and the second insns. If these values are not
6508 acceptable, you could use the hook to modify them too. See also
6509 @pxref{Processor pipeline description}.
6512 @hook TARGET_SCHED_ADJUST_PRIORITY
6513 This hook adjusts the integer scheduling priority @var{priority} of
6514 @var{insn}. It should return the new priority. Increase the priority to
6515 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6516 later. Do not define this hook if you do not need to adjust the
6517 scheduling priorities of insns.
6520 @hook TARGET_SCHED_REORDER
6521 This hook is executed by the scheduler after it has scheduled the ready
6522 list, to allow the machine description to reorder it (for example to
6523 combine two small instructions together on @samp{VLIW} machines).
6524 @var{file} is either a null pointer, or a stdio stream to write any
6525 debug output to. @var{verbose} is the verbose level provided by
6526 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6527 list of instructions that are ready to be scheduled. @var{n_readyp} is
6528 a pointer to the number of elements in the ready list. The scheduler
6529 reads the ready list in reverse order, starting with
6530 @var{ready}[@var{*n_readyp} @minus{} 1] and going to @var{ready}[0]. @var{clock}
6531 is the timer tick of the scheduler. You may modify the ready list and
6532 the number of ready insns. The return value is the number of insns that
6533 can issue this cycle; normally this is just @code{issue_rate}. See also
6534 @samp{TARGET_SCHED_REORDER2}.
6537 @hook TARGET_SCHED_REORDER2
6538 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6539 function is called whenever the scheduler starts a new cycle. This one
6540 is called once per iteration over a cycle, immediately after
6541 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6542 return the number of insns to be scheduled in the same cycle. Defining
6543 this hook can be useful if there are frequent situations where
6544 scheduling one insn causes other insns to become ready in the same
6545 cycle. These other insns can then be taken into account properly.
6548 @hook TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK
6549 This hook is called after evaluation forward dependencies of insns in
6550 chain given by two parameter values (@var{head} and @var{tail}
6551 correspondingly) but before insns scheduling of the insn chain. For
6552 example, it can be used for better insn classification if it requires
6553 analysis of dependencies. This hook can use backward and forward
6554 dependencies of the insn scheduler because they are already
6558 @hook TARGET_SCHED_INIT
6559 This hook is executed by the scheduler at the beginning of each block of
6560 instructions that are to be scheduled. @var{file} is either a null
6561 pointer, or a stdio stream to write any debug output to. @var{verbose}
6562 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6563 @var{max_ready} is the maximum number of insns in the current scheduling
6564 region that can be live at the same time. This can be used to allocate
6565 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6568 @hook TARGET_SCHED_FINISH
6569 This hook is executed by the scheduler at the end of each block of
6570 instructions that are to be scheduled. It can be used to perform
6571 cleanup of any actions done by the other scheduling hooks. @var{file}
6572 is either a null pointer, or a stdio stream to write any debug output
6573 to. @var{verbose} is the verbose level provided by
6574 @option{-fsched-verbose-@var{n}}.
6577 @hook TARGET_SCHED_INIT_GLOBAL
6578 This hook is executed by the scheduler after function level initializations.
6579 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6580 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6581 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6584 @hook TARGET_SCHED_FINISH_GLOBAL
6585 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6586 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6587 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6590 @hook TARGET_SCHED_DFA_PRE_CYCLE_INSN
6591 The hook returns an RTL insn. The automaton state used in the
6592 pipeline hazard recognizer is changed as if the insn were scheduled
6593 when the new simulated processor cycle starts. Usage of the hook may
6594 simplify the automaton pipeline description for some @acronym{VLIW}
6595 processors. If the hook is defined, it is used only for the automaton
6596 based pipeline description. The default is not to change the state
6597 when the new simulated processor cycle starts.
6600 @hook TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN
6601 The hook can be used to initialize data used by the previous hook.
6604 @hook TARGET_SCHED_DFA_POST_CYCLE_INSN
6605 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6606 to changed the state as if the insn were scheduled when the new
6607 simulated processor cycle finishes.
6610 @hook TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN
6611 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6612 used to initialize data used by the previous hook.
6615 @hook TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE
6616 The hook to notify target that the current simulated cycle is about to finish.
6617 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6618 to change the state in more complicated situations - e.g., when advancing
6619 state on a single insn is not enough.
6622 @hook TARGET_SCHED_DFA_POST_ADVANCE_CYCLE
6623 The hook to notify target that new simulated cycle has just started.
6624 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6625 to change the state in more complicated situations - e.g., when advancing
6626 state on a single insn is not enough.
6629 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD
6630 This hook controls better choosing an insn from the ready insn queue
6631 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6632 chooses the first insn from the queue. If the hook returns a positive
6633 value, an additional scheduler code tries all permutations of
6634 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6635 subsequent ready insns to choose an insn whose issue will result in
6636 maximal number of issued insns on the same cycle. For the
6637 @acronym{VLIW} processor, the code could actually solve the problem of
6638 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6639 rules of @acronym{VLIW} packing are described in the automaton.
6641 This code also could be used for superscalar @acronym{RISC}
6642 processors. Let us consider a superscalar @acronym{RISC} processor
6643 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6644 @var{B}, some insns can be executed only in pipelines @var{B} or
6645 @var{C}, and one insn can be executed in pipeline @var{B}. The
6646 processor may issue the 1st insn into @var{A} and the 2nd one into
6647 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6648 until the next cycle. If the scheduler issues the 3rd insn the first,
6649 the processor could issue all 3 insns per cycle.
6651 Actually this code demonstrates advantages of the automaton based
6652 pipeline hazard recognizer. We try quickly and easy many insn
6653 schedules to choose the best one.
6655 The default is no multipass scheduling.
6658 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD
6660 This hook controls what insns from the ready insn queue will be
6661 considered for the multipass insn scheduling. If the hook returns
6662 zero for @var{insn}, the insn will be not chosen to
6665 The default is that any ready insns can be chosen to be issued.
6668 @hook TARGET_SCHED_DFA_NEW_CYCLE
6669 This hook is called by the insn scheduler before issuing @var{insn}
6670 on cycle @var{clock}. If the hook returns nonzero,
6671 @var{insn} is not issued on this processor cycle. Instead,
6672 the processor cycle is advanced. If *@var{sort_p}
6673 is zero, the insn ready queue is not sorted on the new cycle
6674 start as usually. @var{dump} and @var{verbose} specify the file and
6675 verbosity level to use for debugging output.
6676 @var{last_clock} and @var{clock} are, respectively, the
6677 processor cycle on which the previous insn has been issued,
6678 and the current processor cycle.
6681 @hook TARGET_SCHED_IS_COSTLY_DEPENDENCE
6682 This hook is used to define which dependences are considered costly by
6683 the target, so costly that it is not advisable to schedule the insns that
6684 are involved in the dependence too close to one another. The parameters
6685 to this hook are as follows: The first parameter @var{_dep} is the dependence
6686 being evaluated. The second parameter @var{cost} is the cost of the
6687 dependence as estimated by the scheduler, and the third
6688 parameter @var{distance} is the distance in cycles between the two insns.
6689 The hook returns @code{true} if considering the distance between the two
6690 insns the dependence between them is considered costly by the target,
6691 and @code{false} otherwise.
6693 Defining this hook can be useful in multiple-issue out-of-order machines,
6694 where (a) it's practically hopeless to predict the actual data/resource
6695 delays, however: (b) there's a better chance to predict the actual grouping
6696 that will be formed, and (c) correctly emulating the grouping can be very
6697 important. In such targets one may want to allow issuing dependent insns
6698 closer to one another---i.e., closer than the dependence distance; however,
6699 not in cases of ``costly dependences'', which this hooks allows to define.
6702 @hook TARGET_SCHED_H_I_D_EXTENDED
6703 This hook is called by the insn scheduler after emitting a new instruction to
6704 the instruction stream. The hook notifies a target backend to extend its
6705 per instruction data structures.
6708 @hook TARGET_SCHED_ALLOC_SCHED_CONTEXT
6709 Return a pointer to a store large enough to hold target scheduling context.
6712 @hook TARGET_SCHED_INIT_SCHED_CONTEXT
6713 Initialize store pointed to by @var{tc} to hold target scheduling context.
6714 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6715 beginning of the block. Otherwise, copy the current context into @var{tc}.
6718 @hook TARGET_SCHED_SET_SCHED_CONTEXT
6719 Copy target scheduling context pointed to by @var{tc} to the current context.
6722 @hook TARGET_SCHED_CLEAR_SCHED_CONTEXT
6723 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6726 @hook TARGET_SCHED_FREE_SCHED_CONTEXT
6727 Deallocate a store for target scheduling context pointed to by @var{tc}.
6730 @hook TARGET_SCHED_SPECULATE_INSN
6731 This hook is called by the insn scheduler when @var{insn} has only
6732 speculative dependencies and therefore can be scheduled speculatively.
6733 The hook is used to check if the pattern of @var{insn} has a speculative
6734 version and, in case of successful check, to generate that speculative
6735 pattern. The hook should return 1, if the instruction has a speculative form,
6736 or @minus{}1, if it doesn't. @var{request} describes the type of requested
6737 speculation. If the return value equals 1 then @var{new_pat} is assigned
6738 the generated speculative pattern.
6741 @hook TARGET_SCHED_NEEDS_BLOCK_P
6742 This hook is called by the insn scheduler during generation of recovery code
6743 for @var{insn}. It should return @code{true}, if the corresponding check
6744 instruction should branch to recovery code, or @code{false} otherwise.
6747 @hook TARGET_SCHED_GEN_SPEC_CHECK
6748 This hook is called by the insn scheduler to generate a pattern for recovery
6749 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6750 speculative instruction for which the check should be generated.
6751 @var{label} is either a label of a basic block, where recovery code should
6752 be emitted, or a null pointer, when requested check doesn't branch to
6753 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6754 a pattern for a branchy check corresponding to a simple check denoted by
6755 @var{insn} should be generated. In this case @var{label} can't be null.
6758 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC
6759 This hook is used as a workaround for
6760 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6761 called on the first instruction of the ready list. The hook is used to
6762 discard speculative instructions that stand first in the ready list from
6763 being scheduled on the current cycle. If the hook returns @code{false},
6764 @var{insn} will not be chosen to be issued.
6765 For non-speculative instructions,
6766 the hook should always return @code{true}. For example, in the ia64 backend
6767 the hook is used to cancel data speculative insns when the ALAT table
6771 @hook TARGET_SCHED_SET_SCHED_FLAGS
6772 This hook is used by the insn scheduler to find out what features should be
6774 The structure *@var{spec_info} should be filled in by the target.
6775 The structure describes speculation types that can be used in the scheduler.
6778 @hook TARGET_SCHED_SMS_RES_MII
6779 This hook is called by the swing modulo scheduler to calculate a
6780 resource-based lower bound which is based on the resources available in
6781 the machine and the resources required by each instruction. The target
6782 backend can use @var{g} to calculate such bound. A very simple lower
6783 bound will be used in case this hook is not implemented: the total number
6784 of instructions divided by the issue rate.
6788 @section Dividing the Output into Sections (Texts, Data, @dots{})
6789 @c the above section title is WAY too long. maybe cut the part between
6790 @c the (...)? --mew 10feb93
6792 An object file is divided into sections containing different types of
6793 data. In the most common case, there are three sections: the @dfn{text
6794 section}, which holds instructions and read-only data; the @dfn{data
6795 section}, which holds initialized writable data; and the @dfn{bss
6796 section}, which holds uninitialized data. Some systems have other kinds
6799 @file{varasm.c} provides several well-known sections, such as
6800 @code{text_section}, @code{data_section} and @code{bss_section}.
6801 The normal way of controlling a @code{@var{foo}_section} variable
6802 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6803 as described below. The macros are only read once, when @file{varasm.c}
6804 initializes itself, so their values must be run-time constants.
6805 They may however depend on command-line flags.
6807 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6808 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6809 to be string literals.
6811 Some assemblers require a different string to be written every time a
6812 section is selected. If your assembler falls into this category, you
6813 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6814 @code{get_unnamed_section} to set up the sections.
6816 You must always create a @code{text_section}, either by defining
6817 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6818 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6819 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6820 create a distinct @code{readonly_data_section}, the default is to
6821 reuse @code{text_section}.
6823 All the other @file{varasm.c} sections are optional, and are null
6824 if the target does not provide them.
6826 @defmac TEXT_SECTION_ASM_OP
6827 A C expression whose value is a string, including spacing, containing the
6828 assembler operation that should precede instructions and read-only data.
6829 Normally @code{"\t.text"} is right.
6832 @defmac HOT_TEXT_SECTION_NAME
6833 If defined, a C string constant for the name of the section containing most
6834 frequently executed functions of the program. If not defined, GCC will provide
6835 a default definition if the target supports named sections.
6838 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6839 If defined, a C string constant for the name of the section containing unlikely
6840 executed functions in the program.
6843 @defmac DATA_SECTION_ASM_OP
6844 A C expression whose value is a string, including spacing, containing the
6845 assembler operation to identify the following data as writable initialized
6846 data. Normally @code{"\t.data"} is right.
6849 @defmac SDATA_SECTION_ASM_OP
6850 If defined, a C expression whose value is a string, including spacing,
6851 containing the assembler operation to identify the following data as
6852 initialized, writable small data.
6855 @defmac READONLY_DATA_SECTION_ASM_OP
6856 A C expression whose value is a string, including spacing, containing the
6857 assembler operation to identify the following data as read-only initialized
6861 @defmac BSS_SECTION_ASM_OP
6862 If defined, a C expression whose value is a string, including spacing,
6863 containing the assembler operation to identify the following data as
6864 uninitialized global data. If not defined, and neither
6865 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
6866 uninitialized global data will be output in the data section if
6867 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6871 @defmac SBSS_SECTION_ASM_OP
6872 If defined, a C expression whose value is a string, including spacing,
6873 containing the assembler operation to identify the following data as
6874 uninitialized, writable small data.
6877 @defmac TLS_COMMON_ASM_OP
6878 If defined, a C expression whose value is a string containing the
6879 assembler operation to identify the following data as thread-local
6880 common data. The default is @code{".tls_common"}.
6883 @defmac TLS_SECTION_ASM_FLAG
6884 If defined, a C expression whose value is a character constant
6885 containing the flag used to mark a section as a TLS section. The
6886 default is @code{'T'}.
6889 @defmac INIT_SECTION_ASM_OP
6890 If defined, a C expression whose value is a string, including spacing,
6891 containing the assembler operation to identify the following data as
6892 initialization code. If not defined, GCC will assume such a section does
6893 not exist. This section has no corresponding @code{init_section}
6894 variable; it is used entirely in runtime code.
6897 @defmac FINI_SECTION_ASM_OP
6898 If defined, a C expression whose value is a string, including spacing,
6899 containing the assembler operation to identify the following data as
6900 finalization code. If not defined, GCC will assume such a section does
6901 not exist. This section has no corresponding @code{fini_section}
6902 variable; it is used entirely in runtime code.
6905 @defmac INIT_ARRAY_SECTION_ASM_OP
6906 If defined, a C expression whose value is a string, including spacing,
6907 containing the assembler operation to identify the following data as
6908 part of the @code{.init_array} (or equivalent) section. If not
6909 defined, GCC will assume such a section does not exist. Do not define
6910 both this macro and @code{INIT_SECTION_ASM_OP}.
6913 @defmac FINI_ARRAY_SECTION_ASM_OP
6914 If defined, a C expression whose value is a string, including spacing,
6915 containing the assembler operation to identify the following data as
6916 part of the @code{.fini_array} (or equivalent) section. If not
6917 defined, GCC will assume such a section does not exist. Do not define
6918 both this macro and @code{FINI_SECTION_ASM_OP}.
6921 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6922 If defined, an ASM statement that switches to a different section
6923 via @var{section_op}, calls @var{function}, and switches back to
6924 the text section. This is used in @file{crtstuff.c} if
6925 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6926 to initialization and finalization functions from the init and fini
6927 sections. By default, this macro uses a simple function call. Some
6928 ports need hand-crafted assembly code to avoid dependencies on
6929 registers initialized in the function prologue or to ensure that
6930 constant pools don't end up too far way in the text section.
6933 @defmac TARGET_LIBGCC_SDATA_SECTION
6934 If defined, a string which names the section into which small
6935 variables defined in crtstuff and libgcc should go. This is useful
6936 when the target has options for optimizing access to small data, and
6937 you want the crtstuff and libgcc routines to be conservative in what
6938 they expect of your application yet liberal in what your application
6939 expects. For example, for targets with a @code{.sdata} section (like
6940 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6941 require small data support from your application, but use this macro
6942 to put small data into @code{.sdata} so that your application can
6943 access these variables whether it uses small data or not.
6946 @defmac FORCE_CODE_SECTION_ALIGN
6947 If defined, an ASM statement that aligns a code section to some
6948 arbitrary boundary. This is used to force all fragments of the
6949 @code{.init} and @code{.fini} sections to have to same alignment
6950 and thus prevent the linker from having to add any padding.
6953 @defmac JUMP_TABLES_IN_TEXT_SECTION
6954 Define this macro to be an expression with a nonzero value if jump
6955 tables (for @code{tablejump} insns) should be output in the text
6956 section, along with the assembler instructions. Otherwise, the
6957 readonly data section is used.
6959 This macro is irrelevant if there is no separate readonly data section.
6962 @hook TARGET_ASM_INIT_SECTIONS
6963 Define this hook if you need to do something special to set up the
6964 @file{varasm.c} sections, or if your target has some special sections
6965 of its own that you need to create.
6967 GCC calls this hook after processing the command line, but before writing
6968 any assembly code, and before calling any of the section-returning hooks
6972 @hook TARGET_ASM_RELOC_RW_MASK
6973 Return a mask describing how relocations should be treated when
6974 selecting sections. Bit 1 should be set if global relocations
6975 should be placed in a read-write section; bit 0 should be set if
6976 local relocations should be placed in a read-write section.
6978 The default version of this function returns 3 when @option{-fpic}
6979 is in effect, and 0 otherwise. The hook is typically redefined
6980 when the target cannot support (some kinds of) dynamic relocations
6981 in read-only sections even in executables.
6984 @hook TARGET_ASM_SELECT_SECTION
6985 Return the section into which @var{exp} should be placed. You can
6986 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6987 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6988 requires link-time relocations. Bit 0 is set when variable contains
6989 local relocations only, while bit 1 is set for global relocations.
6990 @var{align} is the constant alignment in bits.
6992 The default version of this function takes care of putting read-only
6993 variables in @code{readonly_data_section}.
6995 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6998 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6999 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
7000 for @code{FUNCTION_DECL}s as well as for variables and constants.
7002 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
7003 function has been determined to be likely to be called, and nonzero if
7004 it is unlikely to be called.
7007 @hook TARGET_ASM_UNIQUE_SECTION
7008 Build up a unique section name, expressed as a @code{STRING_CST} node,
7009 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
7010 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
7011 the initial value of @var{exp} requires link-time relocations.
7013 The default version of this function appends the symbol name to the
7014 ELF section name that would normally be used for the symbol. For
7015 example, the function @code{foo} would be placed in @code{.text.foo}.
7016 Whatever the actual target object format, this is often good enough.
7019 @hook TARGET_ASM_FUNCTION_RODATA_SECTION
7020 Return the readonly data section associated with
7021 @samp{DECL_SECTION_NAME (@var{decl})}.
7022 The default version of this function selects @code{.gnu.linkonce.r.name} if
7023 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
7024 if function is in @code{.text.name}, and the normal readonly-data section
7028 @hook TARGET_ASM_SELECT_RTX_SECTION
7029 Return the section into which a constant @var{x}, of mode @var{mode},
7030 should be placed. You can assume that @var{x} is some kind of
7031 constant in RTL@. The argument @var{mode} is redundant except in the
7032 case of a @code{const_int} rtx. @var{align} is the constant alignment
7035 The default version of this function takes care of putting symbolic
7036 constants in @code{flag_pic} mode in @code{data_section} and everything
7037 else in @code{readonly_data_section}.
7040 @hook TARGET_MANGLE_DECL_ASSEMBLER_NAME
7041 Define this hook if you need to postprocess the assembler name generated
7042 by target-independent code. The @var{id} provided to this hook will be
7043 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
7044 or the mangled name of the @var{decl} in C++). The return value of the
7045 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
7046 your target system. The default implementation of this hook just
7047 returns the @var{id} provided.
7050 @hook TARGET_ENCODE_SECTION_INFO
7051 Define this hook if references to a symbol or a constant must be
7052 treated differently depending on something about the variable or
7053 function named by the symbol (such as what section it is in).
7055 The hook is executed immediately after rtl has been created for
7056 @var{decl}, which may be a variable or function declaration or
7057 an entry in the constant pool. In either case, @var{rtl} is the
7058 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
7059 in this hook; that field may not have been initialized yet.
7061 In the case of a constant, it is safe to assume that the rtl is
7062 a @code{mem} whose address is a @code{symbol_ref}. Most decls
7063 will also have this form, but that is not guaranteed. Global
7064 register variables, for instance, will have a @code{reg} for their
7065 rtl. (Normally the right thing to do with such unusual rtl is
7068 The @var{new_decl_p} argument will be true if this is the first time
7069 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
7070 be false for subsequent invocations, which will happen for duplicate
7071 declarations. Whether or not anything must be done for the duplicate
7072 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
7073 @var{new_decl_p} is always true when the hook is called for a constant.
7075 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
7076 The usual thing for this hook to do is to record flags in the
7077 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
7078 Historically, the name string was modified if it was necessary to
7079 encode more than one bit of information, but this practice is now
7080 discouraged; use @code{SYMBOL_REF_FLAGS}.
7082 The default definition of this hook, @code{default_encode_section_info}
7083 in @file{varasm.c}, sets a number of commonly-useful bits in
7084 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
7085 before overriding it.
7088 @hook TARGET_STRIP_NAME_ENCODING
7089 Decode @var{name} and return the real name part, sans
7090 the characters that @code{TARGET_ENCODE_SECTION_INFO}
7094 @hook TARGET_IN_SMALL_DATA_P
7095 Returns true if @var{exp} should be placed into a ``small data'' section.
7096 The default version of this hook always returns false.
7099 @hook TARGET_HAVE_SRODATA_SECTION
7100 Contains the value true if the target places read-only
7101 ``small data'' into a separate section. The default value is false.
7104 @hook TARGET_BINDS_LOCAL_P
7105 Returns true if @var{exp} names an object for which name resolution
7106 rules must resolve to the current ``module'' (dynamic shared library
7107 or executable image).
7109 The default version of this hook implements the name resolution rules
7110 for ELF, which has a looser model of global name binding than other
7111 currently supported object file formats.
7114 @hook TARGET_HAVE_TLS
7115 Contains the value true if the target supports thread-local storage.
7116 The default value is false.
7121 @section Position Independent Code
7122 @cindex position independent code
7125 This section describes macros that help implement generation of position
7126 independent code. Simply defining these macros is not enough to
7127 generate valid PIC; you must also add support to the hook
7128 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
7129 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
7130 must modify the definition of @samp{movsi} to do something appropriate
7131 when the source operand contains a symbolic address. You may also
7132 need to alter the handling of switch statements so that they use
7134 @c i rearranged the order of the macros above to try to force one of
7135 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
7137 @defmac PIC_OFFSET_TABLE_REGNUM
7138 The register number of the register used to address a table of static
7139 data addresses in memory. In some cases this register is defined by a
7140 processor's ``application binary interface'' (ABI)@. When this macro
7141 is defined, RTL is generated for this register once, as with the stack
7142 pointer and frame pointer registers. If this macro is not defined, it
7143 is up to the machine-dependent files to allocate such a register (if
7144 necessary). Note that this register must be fixed when in use (e.g.@:
7145 when @code{flag_pic} is true).
7148 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7149 Define this macro if the register defined by
7150 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
7151 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7154 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7155 A C expression that is nonzero if @var{x} is a legitimate immediate
7156 operand on the target machine when generating position independent code.
7157 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7158 check this. You can also assume @var{flag_pic} is true, so you need not
7159 check it either. You need not define this macro if all constants
7160 (including @code{SYMBOL_REF}) can be immediate operands when generating
7161 position independent code.
7164 @node Assembler Format
7165 @section Defining the Output Assembler Language
7167 This section describes macros whose principal purpose is to describe how
7168 to write instructions in assembler language---rather than what the
7172 * File Framework:: Structural information for the assembler file.
7173 * Data Output:: Output of constants (numbers, strings, addresses).
7174 * Uninitialized Data:: Output of uninitialized variables.
7175 * Label Output:: Output and generation of labels.
7176 * Initialization:: General principles of initialization
7177 and termination routines.
7178 * Macros for Initialization::
7179 Specific macros that control the handling of
7180 initialization and termination routines.
7181 * Instruction Output:: Output of actual instructions.
7182 * Dispatch Tables:: Output of jump tables.
7183 * Exception Region Output:: Output of exception region code.
7184 * Alignment Output:: Pseudo ops for alignment and skipping data.
7187 @node File Framework
7188 @subsection The Overall Framework of an Assembler File
7189 @cindex assembler format
7190 @cindex output of assembler code
7192 @c prevent bad page break with this line
7193 This describes the overall framework of an assembly file.
7195 @findex default_file_start
7196 @hook TARGET_ASM_FILE_START
7197 Output to @code{asm_out_file} any text which the assembler expects to
7198 find at the beginning of a file. The default behavior is controlled
7199 by two flags, documented below. Unless your target's assembler is
7200 quite unusual, if you override the default, you should call
7201 @code{default_file_start} at some point in your target hook. This
7202 lets other target files rely on these variables.
7205 @hook TARGET_ASM_FILE_START_APP_OFF
7206 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7207 printed as the very first line in the assembly file, unless
7208 @option{-fverbose-asm} is in effect. (If that macro has been defined
7209 to the empty string, this variable has no effect.) With the normal
7210 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7211 assembler that it need not bother stripping comments or extra
7212 whitespace from its input. This allows it to work a bit faster.
7214 The default is false. You should not set it to true unless you have
7215 verified that your port does not generate any extra whitespace or
7216 comments that will cause GAS to issue errors in NO_APP mode.
7219 @hook TARGET_ASM_FILE_START_FILE_DIRECTIVE
7220 If this flag is true, @code{output_file_directive} will be called
7221 for the primary source file, immediately after printing
7222 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7223 this to be done. The default is false.
7226 @hook TARGET_ASM_FILE_END
7227 Output to @code{asm_out_file} any text which the assembler expects
7228 to find at the end of a file. The default is to output nothing.
7231 @deftypefun void file_end_indicate_exec_stack ()
7232 Some systems use a common convention, the @samp{.note.GNU-stack}
7233 special section, to indicate whether or not an object file relies on
7234 the stack being executable. If your system uses this convention, you
7235 should define @code{TARGET_ASM_FILE_END} to this function. If you
7236 need to do other things in that hook, have your hook function call
7240 @hook TARGET_ASM_LTO_START
7241 Output to @code{asm_out_file} any text which the assembler expects
7242 to find at the start of an LTO section. The default is to output
7246 @hook TARGET_ASM_LTO_END
7247 Output to @code{asm_out_file} any text which the assembler expects
7248 to find at the end of an LTO section. The default is to output
7252 @hook TARGET_ASM_CODE_END
7253 Output to @code{asm_out_file} any text which is needed before emitting
7254 unwind info and debug info at the end of a file. Some targets emit
7255 here PIC setup thunks that cannot be emitted at the end of file,
7256 because they couldn't have unwind info then. The default is to output
7260 @defmac ASM_COMMENT_START
7261 A C string constant describing how to begin a comment in the target
7262 assembler language. The compiler assumes that the comment will end at
7263 the end of the line.
7267 A C string constant for text to be output before each @code{asm}
7268 statement or group of consecutive ones. Normally this is
7269 @code{"#APP"}, which is a comment that has no effect on most
7270 assemblers but tells the GNU assembler that it must check the lines
7271 that follow for all valid assembler constructs.
7275 A C string constant for text to be output after each @code{asm}
7276 statement or group of consecutive ones. Normally this is
7277 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7278 time-saving assumptions that are valid for ordinary compiler output.
7281 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7282 A C statement to output COFF information or DWARF debugging information
7283 which indicates that filename @var{name} is the current source file to
7284 the stdio stream @var{stream}.
7286 This macro need not be defined if the standard form of output
7287 for the file format in use is appropriate.
7290 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7291 A C statement to output the string @var{string} to the stdio stream
7292 @var{stream}. If you do not call the function @code{output_quoted_string}
7293 in your config files, GCC will only call it to output filenames to
7294 the assembler source. So you can use it to canonicalize the format
7295 of the filename using this macro.
7298 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
7299 A C statement to output something to the assembler file to handle a
7300 @samp{#ident} directive containing the text @var{string}. If this
7301 macro is not defined, nothing is output for a @samp{#ident} directive.
7304 @hook TARGET_ASM_NAMED_SECTION
7305 Output assembly directives to switch to section @var{name}. The section
7306 should have attributes as specified by @var{flags}, which is a bit mask
7307 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{decl}
7308 is non-NULL, it is the @code{VAR_DECL} or @code{FUNCTION_DECL} with which
7309 this section is associated.
7312 @hook TARGET_HAVE_NAMED_SECTIONS
7313 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7316 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7317 @hook TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7318 This flag is true if we can create zeroed data by switching to a BSS
7319 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7320 This is true on most ELF targets.
7323 @hook TARGET_SECTION_TYPE_FLAGS
7324 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7325 based on a variable or function decl, a section name, and whether or not the
7326 declaration's initializer may contain runtime relocations. @var{decl} may be
7327 null, in which case read-write data should be assumed.
7329 The default version of this function handles choosing code vs data,
7330 read-only vs read-write data, and @code{flag_pic}. You should only
7331 need to override this if your target has special flags that might be
7332 set via @code{__attribute__}.
7335 @hook TARGET_ASM_RECORD_GCC_SWITCHES
7336 Provides the target with the ability to record the gcc command line
7337 switches that have been passed to the compiler, and options that are
7338 enabled. The @var{type} argument specifies what is being recorded.
7339 It can take the following values:
7342 @item SWITCH_TYPE_PASSED
7343 @var{text} is a command line switch that has been set by the user.
7345 @item SWITCH_TYPE_ENABLED
7346 @var{text} is an option which has been enabled. This might be as a
7347 direct result of a command line switch, or because it is enabled by
7348 default or because it has been enabled as a side effect of a different
7349 command line switch. For example, the @option{-O2} switch enables
7350 various different individual optimization passes.
7352 @item SWITCH_TYPE_DESCRIPTIVE
7353 @var{text} is either NULL or some descriptive text which should be
7354 ignored. If @var{text} is NULL then it is being used to warn the
7355 target hook that either recording is starting or ending. The first
7356 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7357 warning is for start up and the second time the warning is for
7358 wind down. This feature is to allow the target hook to make any
7359 necessary preparations before it starts to record switches and to
7360 perform any necessary tidying up after it has finished recording
7363 @item SWITCH_TYPE_LINE_START
7364 This option can be ignored by this target hook.
7366 @item SWITCH_TYPE_LINE_END
7367 This option can be ignored by this target hook.
7370 The hook's return value must be zero. Other return values may be
7371 supported in the future.
7373 By default this hook is set to NULL, but an example implementation is
7374 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7375 it records the switches as ASCII text inside a new, string mergeable
7376 section in the assembler output file. The name of the new section is
7377 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7381 @hook TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7382 This is the name of the section that will be created by the example
7383 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7389 @subsection Output of Data
7392 @hook TARGET_ASM_BYTE_OP
7393 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7394 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7395 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7396 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7397 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7398 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7399 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7400 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7401 These hooks specify assembly directives for creating certain kinds
7402 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7403 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7404 aligned two-byte object, and so on. Any of the hooks may be
7405 @code{NULL}, indicating that no suitable directive is available.
7407 The compiler will print these strings at the start of a new line,
7408 followed immediately by the object's initial value. In most cases,
7409 the string should contain a tab, a pseudo-op, and then another tab.
7412 @hook TARGET_ASM_INTEGER
7413 The @code{assemble_integer} function uses this hook to output an
7414 integer object. @var{x} is the object's value, @var{size} is its size
7415 in bytes and @var{aligned_p} indicates whether it is aligned. The
7416 function should return @code{true} if it was able to output the
7417 object. If it returns false, @code{assemble_integer} will try to
7418 split the object into smaller parts.
7420 The default implementation of this hook will use the
7421 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7422 when the relevant string is @code{NULL}.
7425 @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
7426 A C statement to recognize @var{rtx} patterns that
7427 @code{output_addr_const} can't deal with, and output assembly code to
7428 @var{stream} corresponding to the pattern @var{x}. This may be used to
7429 allow machine-dependent @code{UNSPEC}s to appear within constants.
7431 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
7432 @code{goto fail}, so that a standard error message is printed. If it
7433 prints an error message itself, by calling, for example,
7434 @code{output_operand_lossage}, it may just complete normally.
7437 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7438 A C statement to output to the stdio stream @var{stream} an assembler
7439 instruction to assemble a string constant containing the @var{len}
7440 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7441 @code{char *} and @var{len} a C expression of type @code{int}.
7443 If the assembler has a @code{.ascii} pseudo-op as found in the
7444 Berkeley Unix assembler, do not define the macro
7445 @code{ASM_OUTPUT_ASCII}.
7448 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7449 A C statement to output word @var{n} of a function descriptor for
7450 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7451 is defined, and is otherwise unused.
7454 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7455 You may define this macro as a C expression. You should define the
7456 expression to have a nonzero value if GCC should output the constant
7457 pool for a function before the code for the function, or a zero value if
7458 GCC should output the constant pool after the function. If you do
7459 not define this macro, the usual case, GCC will output the constant
7460 pool before the function.
7463 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7464 A C statement to output assembler commands to define the start of the
7465 constant pool for a function. @var{funname} is a string giving
7466 the name of the function. Should the return type of the function
7467 be required, it can be obtained via @var{fundecl}. @var{size}
7468 is the size, in bytes, of the constant pool that will be written
7469 immediately after this call.
7471 If no constant-pool prefix is required, the usual case, this macro need
7475 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7476 A C statement (with or without semicolon) to output a constant in the
7477 constant pool, if it needs special treatment. (This macro need not do
7478 anything for RTL expressions that can be output normally.)
7480 The argument @var{file} is the standard I/O stream to output the
7481 assembler code on. @var{x} is the RTL expression for the constant to
7482 output, and @var{mode} is the machine mode (in case @var{x} is a
7483 @samp{const_int}). @var{align} is the required alignment for the value
7484 @var{x}; you should output an assembler directive to force this much
7487 The argument @var{labelno} is a number to use in an internal label for
7488 the address of this pool entry. The definition of this macro is
7489 responsible for outputting the label definition at the proper place.
7490 Here is how to do this:
7493 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7496 When you output a pool entry specially, you should end with a
7497 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7498 entry from being output a second time in the usual manner.
7500 You need not define this macro if it would do nothing.
7503 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7504 A C statement to output assembler commands to at the end of the constant
7505 pool for a function. @var{funname} is a string giving the name of the
7506 function. Should the return type of the function be required, you can
7507 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7508 constant pool that GCC wrote immediately before this call.
7510 If no constant-pool epilogue is required, the usual case, you need not
7514 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7515 Define this macro as a C expression which is nonzero if @var{C} is
7516 used as a logical line separator by the assembler. @var{STR} points
7517 to the position in the string where @var{C} was found; this can be used if
7518 a line separator uses multiple characters.
7520 If you do not define this macro, the default is that only
7521 the character @samp{;} is treated as a logical line separator.
7524 @hook TARGET_ASM_OPEN_PAREN
7525 These target hooks are C string constants, describing the syntax in the
7526 assembler for grouping arithmetic expressions. If not overridden, they
7527 default to normal parentheses, which is correct for most assemblers.
7530 These macros are provided by @file{real.h} for writing the definitions
7531 of @code{ASM_OUTPUT_DOUBLE} and the like:
7533 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7534 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7535 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7536 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7537 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7538 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7539 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7540 target's floating point representation, and store its bit pattern in
7541 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7542 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7543 simple @code{long int}. For the others, it should be an array of
7544 @code{long int}. The number of elements in this array is determined
7545 by the size of the desired target floating point data type: 32 bits of
7546 it go in each @code{long int} array element. Each array element holds
7547 32 bits of the result, even if @code{long int} is wider than 32 bits
7548 on the host machine.
7550 The array element values are designed so that you can print them out
7551 using @code{fprintf} in the order they should appear in the target
7555 @node Uninitialized Data
7556 @subsection Output of Uninitialized Variables
7558 Each of the macros in this section is used to do the whole job of
7559 outputting a single uninitialized variable.
7561 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7562 A C statement (sans semicolon) to output to the stdio stream
7563 @var{stream} the assembler definition of a common-label named
7564 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7565 is the size rounded up to whatever alignment the caller wants. It is
7566 possible that @var{size} may be zero, for instance if a struct with no
7567 other member than a zero-length array is defined. In this case, the
7568 backend must output a symbol definition that allocates at least one
7569 byte, both so that the address of the resulting object does not compare
7570 equal to any other, and because some object formats cannot even express
7571 the concept of a zero-sized common symbol, as that is how they represent
7572 an ordinary undefined external.
7574 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7575 output the name itself; before and after that, output the additional
7576 assembler syntax for defining the name, and a newline.
7578 This macro controls how the assembler definitions of uninitialized
7579 common global variables are output.
7582 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7583 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7584 separate, explicit argument. If you define this macro, it is used in
7585 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7586 handling the required alignment of the variable. The alignment is specified
7587 as the number of bits.
7590 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7591 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7592 variable to be output, if there is one, or @code{NULL_TREE} if there
7593 is no corresponding variable. If you define this macro, GCC will use it
7594 in place of both @code{ASM_OUTPUT_COMMON} and
7595 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7596 the variable's decl in order to chose what to output.
7599 @defmac ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
7600 A C statement (sans semicolon) to output to the stdio stream
7601 @var{stream} the assembler definition of uninitialized global @var{decl} named
7602 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7603 is the size rounded up to whatever alignment the caller wants.
7605 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
7606 defining this macro. If unable, use the expression
7607 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7608 before and after that, output the additional assembler syntax for defining
7609 the name, and a newline.
7611 There are two ways of handling global BSS@. One is to define either
7612 this macro or its aligned counterpart, @code{ASM_OUTPUT_ALIGNED_BSS}.
7613 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7614 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7615 You do not need to do both.
7617 Some languages do not have @code{common} data, and require a
7618 non-common form of global BSS in order to handle uninitialized globals
7619 efficiently. C++ is one example of this. However, if the target does
7620 not support global BSS, the front end may choose to make globals
7621 common in order to save space in the object file.
7624 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7625 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
7626 separate, explicit argument. If you define this macro, it is used in
7627 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
7628 handling the required alignment of the variable. The alignment is specified
7629 as the number of bits.
7631 Try to use function @code{asm_output_aligned_bss} defined in file
7632 @file{varasm.c} when defining this macro.
7635 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7636 A C statement (sans semicolon) to output to the stdio stream
7637 @var{stream} the assembler definition of a local-common-label named
7638 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7639 is the size rounded up to whatever alignment the caller wants.
7641 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7642 output the name itself; before and after that, output the additional
7643 assembler syntax for defining the name, and a newline.
7645 This macro controls how the assembler definitions of uninitialized
7646 static variables are output.
7649 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7650 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7651 separate, explicit argument. If you define this macro, it is used in
7652 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7653 handling the required alignment of the variable. The alignment is specified
7654 as the number of bits.
7657 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7658 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7659 variable to be output, if there is one, or @code{NULL_TREE} if there
7660 is no corresponding variable. If you define this macro, GCC will use it
7661 in place of both @code{ASM_OUTPUT_DECL} and
7662 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7663 the variable's decl in order to chose what to output.
7667 @subsection Output and Generation of Labels
7669 @c prevent bad page break with this line
7670 This is about outputting labels.
7672 @findex assemble_name
7673 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7674 A C statement (sans semicolon) to output to the stdio stream
7675 @var{stream} the assembler definition of a label named @var{name}.
7676 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7677 output the name itself; before and after that, output the additional
7678 assembler syntax for defining the name, and a newline. A default
7679 definition of this macro is provided which is correct for most systems.
7682 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
7683 A C statement (sans semicolon) to output to the stdio stream
7684 @var{stream} the assembler definition of a label named @var{name} of
7686 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7687 output the name itself; before and after that, output the additional
7688 assembler syntax for defining the name, and a newline. A default
7689 definition of this macro is provided which is correct for most systems.
7691 If this macro is not defined, then the function name is defined in the
7692 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7695 @findex assemble_name_raw
7696 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7697 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7698 to refer to a compiler-generated label. The default definition uses
7699 @code{assemble_name_raw}, which is like @code{assemble_name} except
7700 that it is more efficient.
7704 A C string containing the appropriate assembler directive to specify the
7705 size of a symbol, without any arguments. On systems that use ELF, the
7706 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7707 systems, the default is not to define this macro.
7709 Define this macro only if it is correct to use the default definitions
7710 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7711 for your system. If you need your own custom definitions of those
7712 macros, or if you do not need explicit symbol sizes at all, do not
7716 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7717 A C statement (sans semicolon) to output to the stdio stream
7718 @var{stream} a directive telling the assembler that the size of the
7719 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7720 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7724 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7725 A C statement (sans semicolon) to output to the stdio stream
7726 @var{stream} a directive telling the assembler to calculate the size of
7727 the symbol @var{name} by subtracting its address from the current
7730 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7731 provided. The default assumes that the assembler recognizes a special
7732 @samp{.} symbol as referring to the current address, and can calculate
7733 the difference between this and another symbol. If your assembler does
7734 not recognize @samp{.} or cannot do calculations with it, you will need
7735 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7739 A C string containing the appropriate assembler directive to specify the
7740 type of a symbol, without any arguments. On systems that use ELF, the
7741 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7742 systems, the default is not to define this macro.
7744 Define this macro only if it is correct to use the default definition of
7745 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7746 custom definition of this macro, or if you do not need explicit symbol
7747 types at all, do not define this macro.
7750 @defmac TYPE_OPERAND_FMT
7751 A C string which specifies (using @code{printf} syntax) the format of
7752 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7753 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7754 the default is not to define this macro.
7756 Define this macro only if it is correct to use the default definition of
7757 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7758 custom definition of this macro, or if you do not need explicit symbol
7759 types at all, do not define this macro.
7762 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7763 A C statement (sans semicolon) to output to the stdio stream
7764 @var{stream} a directive telling the assembler that the type of the
7765 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7766 that string is always either @samp{"function"} or @samp{"object"}, but
7767 you should not count on this.
7769 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7770 definition of this macro is provided.
7773 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7774 A C statement (sans semicolon) to output to the stdio stream
7775 @var{stream} any text necessary for declaring the name @var{name} of a
7776 function which is being defined. This macro is responsible for
7777 outputting the label definition (perhaps using
7778 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
7779 @code{FUNCTION_DECL} tree node representing the function.
7781 If this macro is not defined, then the function name is defined in the
7782 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
7784 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7788 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7789 A C statement (sans semicolon) to output to the stdio stream
7790 @var{stream} any text necessary for declaring the size of a function
7791 which is being defined. The argument @var{name} is the name of the
7792 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7793 representing the function.
7795 If this macro is not defined, then the function size is not defined.
7797 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7801 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7802 A C statement (sans semicolon) to output to the stdio stream
7803 @var{stream} any text necessary for declaring the name @var{name} of an
7804 initialized variable which is being defined. This macro must output the
7805 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7806 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7808 If this macro is not defined, then the variable name is defined in the
7809 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7811 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7812 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7815 @hook TARGET_ASM_DECLARE_CONSTANT_NAME
7816 A target hook to output to the stdio stream @var{file} any text necessary
7817 for declaring the name @var{name} of a constant which is being defined. This
7818 target hook is responsible for outputting the label definition (perhaps using
7819 @code{assemble_label}). The argument @var{exp} is the value of the constant,
7820 and @var{size} is the size of the constant in bytes. The @var{name}
7821 will be an internal label.
7823 The default version of this target hook, define the @var{name} in the
7824 usual manner as a label (by means of @code{assemble_label}).
7826 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in this target hook.
7829 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7830 A C statement (sans semicolon) to output to the stdio stream
7831 @var{stream} any text necessary for claiming a register @var{regno}
7832 for a global variable @var{decl} with name @var{name}.
7834 If you don't define this macro, that is equivalent to defining it to do
7838 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7839 A C statement (sans semicolon) to finish up declaring a variable name
7840 once the compiler has processed its initializer fully and thus has had a
7841 chance to determine the size of an array when controlled by an
7842 initializer. This is used on systems where it's necessary to declare
7843 something about the size of the object.
7845 If you don't define this macro, that is equivalent to defining it to do
7848 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7849 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7852 @hook TARGET_ASM_GLOBALIZE_LABEL
7853 This target hook is a function to output to the stdio stream
7854 @var{stream} some commands that will make the label @var{name} global;
7855 that is, available for reference from other files.
7857 The default implementation relies on a proper definition of
7858 @code{GLOBAL_ASM_OP}.
7861 @hook TARGET_ASM_GLOBALIZE_DECL_NAME
7862 This target hook is a function to output to the stdio stream
7863 @var{stream} some commands that will make the name associated with @var{decl}
7864 global; that is, available for reference from other files.
7866 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7869 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7870 A C statement (sans semicolon) to output to the stdio stream
7871 @var{stream} some commands that will make the label @var{name} weak;
7872 that is, available for reference from other files but only used if
7873 no other definition is available. Use the expression
7874 @code{assemble_name (@var{stream}, @var{name})} to output the name
7875 itself; before and after that, output the additional assembler syntax
7876 for making that name weak, and a newline.
7878 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7879 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7883 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7884 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7885 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7886 or variable decl. If @var{value} is not @code{NULL}, this C statement
7887 should output to the stdio stream @var{stream} assembler code which
7888 defines (equates) the weak symbol @var{name} to have the value
7889 @var{value}. If @var{value} is @code{NULL}, it should output commands
7890 to make @var{name} weak.
7893 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7894 Outputs a directive that enables @var{name} to be used to refer to
7895 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7896 declaration of @code{name}.
7899 @defmac SUPPORTS_WEAK
7900 A C expression which evaluates to true if the target supports weak symbols.
7902 If you don't define this macro, @file{defaults.h} provides a default
7903 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7904 is defined, the default definition is @samp{1}; otherwise, it is
7905 @samp{0}. Define this macro if you want to control weak symbol support
7906 with a compiler flag such as @option{-melf}.
7909 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7910 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7911 public symbol such that extra copies in multiple translation units will
7912 be discarded by the linker. Define this macro if your object file
7913 format provides support for this concept, such as the @samp{COMDAT}
7914 section flags in the Microsoft Windows PE/COFF format, and this support
7915 requires changes to @var{decl}, such as putting it in a separate section.
7918 @defmac SUPPORTS_ONE_ONLY
7919 A C expression which evaluates to true if the target supports one-only
7922 If you don't define this macro, @file{varasm.c} provides a default
7923 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7924 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7925 you want to control one-only symbol support with a compiler flag, or if
7926 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7927 be emitted as one-only.
7930 @hook TARGET_ASM_ASSEMBLE_VISIBILITY
7931 This target hook is a function to output to @var{asm_out_file} some
7932 commands that will make the symbol(s) associated with @var{decl} have
7933 hidden, protected or internal visibility as specified by @var{visibility}.
7936 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7937 A C expression that evaluates to true if the target's linker expects
7938 that weak symbols do not appear in a static archive's table of contents.
7939 The default is @code{0}.
7941 Leaving weak symbols out of an archive's table of contents means that,
7942 if a symbol will only have a definition in one translation unit and
7943 will have undefined references from other translation units, that
7944 symbol should not be weak. Defining this macro to be nonzero will
7945 thus have the effect that certain symbols that would normally be weak
7946 (explicit template instantiations, and vtables for polymorphic classes
7947 with noninline key methods) will instead be nonweak.
7949 The C++ ABI requires this macro to be zero. Define this macro for
7950 targets where full C++ ABI compliance is impossible and where linker
7951 restrictions require weak symbols to be left out of a static archive's
7955 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
7956 A C statement (sans semicolon) to output to the stdio stream
7957 @var{stream} any text necessary for declaring the name of an external
7958 symbol named @var{name} which is referenced in this compilation but
7959 not defined. The value of @var{decl} is the tree node for the
7962 This macro need not be defined if it does not need to output anything.
7963 The GNU assembler and most Unix assemblers don't require anything.
7966 @hook TARGET_ASM_EXTERNAL_LIBCALL
7967 This target hook is a function to output to @var{asm_out_file} an assembler
7968 pseudo-op to declare a library function name external. The name of the
7969 library function is given by @var{symref}, which is a @code{symbol_ref}.
7972 @hook TARGET_ASM_MARK_DECL_PRESERVED
7973 This target hook is a function to output to @var{asm_out_file} an assembler
7974 directive to annotate @var{symbol} as used. The Darwin target uses the
7975 .no_dead_code_strip directive.
7978 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
7979 A C statement (sans semicolon) to output to the stdio stream
7980 @var{stream} a reference in assembler syntax to a label named
7981 @var{name}. This should add @samp{_} to the front of the name, if that
7982 is customary on your operating system, as it is in most Berkeley Unix
7983 systems. This macro is used in @code{assemble_name}.
7986 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
7987 A C statement (sans semicolon) to output a reference to
7988 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
7989 will be used to output the name of the symbol. This macro may be used
7990 to modify the way a symbol is referenced depending on information
7991 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
7994 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
7995 A C statement (sans semicolon) to output a reference to @var{buf}, the
7996 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
7997 @code{assemble_name} will be used to output the name of the symbol.
7998 This macro is not used by @code{output_asm_label}, or the @code{%l}
7999 specifier that calls it; the intention is that this macro should be set
8000 when it is necessary to output a label differently when its address is
8004 @hook TARGET_ASM_INTERNAL_LABEL
8005 A function to output to the stdio stream @var{stream} a label whose
8006 name is made from the string @var{prefix} and the number @var{labelno}.
8008 It is absolutely essential that these labels be distinct from the labels
8009 used for user-level functions and variables. Otherwise, certain programs
8010 will have name conflicts with internal labels.
8012 It is desirable to exclude internal labels from the symbol table of the
8013 object file. Most assemblers have a naming convention for labels that
8014 should be excluded; on many systems, the letter @samp{L} at the
8015 beginning of a label has this effect. You should find out what
8016 convention your system uses, and follow it.
8018 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
8021 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
8022 A C statement to output to the stdio stream @var{stream} a debug info
8023 label whose name is made from the string @var{prefix} and the number
8024 @var{num}. This is useful for VLIW targets, where debug info labels
8025 may need to be treated differently than branch target labels. On some
8026 systems, branch target labels must be at the beginning of instruction
8027 bundles, but debug info labels can occur in the middle of instruction
8030 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
8034 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
8035 A C statement to store into the string @var{string} a label whose name
8036 is made from the string @var{prefix} and the number @var{num}.
8038 This string, when output subsequently by @code{assemble_name}, should
8039 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
8040 with the same @var{prefix} and @var{num}.
8042 If the string begins with @samp{*}, then @code{assemble_name} will
8043 output the rest of the string unchanged. It is often convenient for
8044 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
8045 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
8046 to output the string, and may change it. (Of course,
8047 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
8048 you should know what it does on your machine.)
8051 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
8052 A C expression to assign to @var{outvar} (which is a variable of type
8053 @code{char *}) a newly allocated string made from the string
8054 @var{name} and the number @var{number}, with some suitable punctuation
8055 added. Use @code{alloca} to get space for the string.
8057 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
8058 produce an assembler label for an internal static variable whose name is
8059 @var{name}. Therefore, the string must be such as to result in valid
8060 assembler code. The argument @var{number} is different each time this
8061 macro is executed; it prevents conflicts between similarly-named
8062 internal static variables in different scopes.
8064 Ideally this string should not be a valid C identifier, to prevent any
8065 conflict with the user's own symbols. Most assemblers allow periods
8066 or percent signs in assembler symbols; putting at least one of these
8067 between the name and the number will suffice.
8069 If this macro is not defined, a default definition will be provided
8070 which is correct for most systems.
8073 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
8074 A C statement to output to the stdio stream @var{stream} assembler code
8075 which defines (equates) the symbol @var{name} to have the value @var{value}.
8078 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8079 correct for most systems.
8082 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
8083 A C statement to output to the stdio stream @var{stream} assembler code
8084 which defines (equates) the symbol whose tree node is @var{decl_of_name}
8085 to have the value of the tree node @var{decl_of_value}. This macro will
8086 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
8087 the tree nodes are available.
8090 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8091 correct for most systems.
8094 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
8095 A C statement that evaluates to true if the assembler code which defines
8096 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
8097 of the tree node @var{decl_of_value} should be emitted near the end of the
8098 current compilation unit. The default is to not defer output of defines.
8099 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
8100 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
8103 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
8104 A C statement to output to the stdio stream @var{stream} assembler code
8105 which defines (equates) the weak symbol @var{name} to have the value
8106 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
8107 an undefined weak symbol.
8109 Define this macro if the target only supports weak aliases; define
8110 @code{ASM_OUTPUT_DEF} instead if possible.
8113 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
8114 Define this macro to override the default assembler names used for
8115 Objective-C methods.
8117 The default name is a unique method number followed by the name of the
8118 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
8119 the category is also included in the assembler name (e.g.@:
8122 These names are safe on most systems, but make debugging difficult since
8123 the method's selector is not present in the name. Therefore, particular
8124 systems define other ways of computing names.
8126 @var{buf} is an expression of type @code{char *} which gives you a
8127 buffer in which to store the name; its length is as long as
8128 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
8129 50 characters extra.
8131 The argument @var{is_inst} specifies whether the method is an instance
8132 method or a class method; @var{class_name} is the name of the class;
8133 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
8134 in a category); and @var{sel_name} is the name of the selector.
8136 On systems where the assembler can handle quoted names, you can use this
8137 macro to provide more human-readable names.
8140 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
8141 A C statement (sans semicolon) to output to the stdio stream
8142 @var{stream} commands to declare that the label @var{name} is an
8143 Objective-C class reference. This is only needed for targets whose
8144 linkers have special support for NeXT-style runtimes.
8147 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
8148 A C statement (sans semicolon) to output to the stdio stream
8149 @var{stream} commands to declare that the label @var{name} is an
8150 unresolved Objective-C class reference. This is only needed for targets
8151 whose linkers have special support for NeXT-style runtimes.
8154 @node Initialization
8155 @subsection How Initialization Functions Are Handled
8156 @cindex initialization routines
8157 @cindex termination routines
8158 @cindex constructors, output of
8159 @cindex destructors, output of
8161 The compiled code for certain languages includes @dfn{constructors}
8162 (also called @dfn{initialization routines})---functions to initialize
8163 data in the program when the program is started. These functions need
8164 to be called before the program is ``started''---that is to say, before
8165 @code{main} is called.
8167 Compiling some languages generates @dfn{destructors} (also called
8168 @dfn{termination routines}) that should be called when the program
8171 To make the initialization and termination functions work, the compiler
8172 must output something in the assembler code to cause those functions to
8173 be called at the appropriate time. When you port the compiler to a new
8174 system, you need to specify how to do this.
8176 There are two major ways that GCC currently supports the execution of
8177 initialization and termination functions. Each way has two variants.
8178 Much of the structure is common to all four variations.
8180 @findex __CTOR_LIST__
8181 @findex __DTOR_LIST__
8182 The linker must build two lists of these functions---a list of
8183 initialization functions, called @code{__CTOR_LIST__}, and a list of
8184 termination functions, called @code{__DTOR_LIST__}.
8186 Each list always begins with an ignored function pointer (which may hold
8187 0, @minus{}1, or a count of the function pointers after it, depending on
8188 the environment). This is followed by a series of zero or more function
8189 pointers to constructors (or destructors), followed by a function
8190 pointer containing zero.
8192 Depending on the operating system and its executable file format, either
8193 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8194 time and exit time. Constructors are called in reverse order of the
8195 list; destructors in forward order.
8197 The best way to handle static constructors works only for object file
8198 formats which provide arbitrarily-named sections. A section is set
8199 aside for a list of constructors, and another for a list of destructors.
8200 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8201 object file that defines an initialization function also puts a word in
8202 the constructor section to point to that function. The linker
8203 accumulates all these words into one contiguous @samp{.ctors} section.
8204 Termination functions are handled similarly.
8206 This method will be chosen as the default by @file{target-def.h} if
8207 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8208 support arbitrary sections, but does support special designated
8209 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8210 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8212 When arbitrary sections are available, there are two variants, depending
8213 upon how the code in @file{crtstuff.c} is called. On systems that
8214 support a @dfn{.init} section which is executed at program startup,
8215 parts of @file{crtstuff.c} are compiled into that section. The
8216 program is linked by the @command{gcc} driver like this:
8219 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8222 The prologue of a function (@code{__init}) appears in the @code{.init}
8223 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8224 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8225 files are provided by the operating system or by the GNU C library, but
8226 are provided by GCC for a few targets.
8228 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8229 compiled from @file{crtstuff.c}. They contain, among other things, code
8230 fragments within the @code{.init} and @code{.fini} sections that branch
8231 to routines in the @code{.text} section. The linker will pull all parts
8232 of a section together, which results in a complete @code{__init} function
8233 that invokes the routines we need at startup.
8235 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8238 If no init section is available, when GCC compiles any function called
8239 @code{main} (or more accurately, any function designated as a program
8240 entry point by the language front end calling @code{expand_main_function}),
8241 it inserts a procedure call to @code{__main} as the first executable code
8242 after the function prologue. The @code{__main} function is defined
8243 in @file{libgcc2.c} and runs the global constructors.
8245 In file formats that don't support arbitrary sections, there are again
8246 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8247 and an `a.out' format must be used. In this case,
8248 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8249 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8250 and with the address of the void function containing the initialization
8251 code as its value. The GNU linker recognizes this as a request to add
8252 the value to a @dfn{set}; the values are accumulated, and are eventually
8253 placed in the executable as a vector in the format described above, with
8254 a leading (ignored) count and a trailing zero element.
8255 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8256 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8257 the compilation of @code{main} to call @code{__main} as above, starting
8258 the initialization process.
8260 The last variant uses neither arbitrary sections nor the GNU linker.
8261 This is preferable when you want to do dynamic linking and when using
8262 file formats which the GNU linker does not support, such as `ECOFF'@. In
8263 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8264 termination functions are recognized simply by their names. This requires
8265 an extra program in the linkage step, called @command{collect2}. This program
8266 pretends to be the linker, for use with GCC; it does its job by running
8267 the ordinary linker, but also arranges to include the vectors of
8268 initialization and termination functions. These functions are called
8269 via @code{__main} as described above. In order to use this method,
8270 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8273 The following section describes the specific macros that control and
8274 customize the handling of initialization and termination functions.
8277 @node Macros for Initialization
8278 @subsection Macros Controlling Initialization Routines
8280 Here are the macros that control how the compiler handles initialization
8281 and termination functions:
8283 @defmac INIT_SECTION_ASM_OP
8284 If defined, a C string constant, including spacing, for the assembler
8285 operation to identify the following data as initialization code. If not
8286 defined, GCC will assume such a section does not exist. When you are
8287 using special sections for initialization and termination functions, this
8288 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8289 run the initialization functions.
8292 @defmac HAS_INIT_SECTION
8293 If defined, @code{main} will not call @code{__main} as described above.
8294 This macro should be defined for systems that control start-up code
8295 on a symbol-by-symbol basis, such as OSF/1, and should not
8296 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8299 @defmac LD_INIT_SWITCH
8300 If defined, a C string constant for a switch that tells the linker that
8301 the following symbol is an initialization routine.
8304 @defmac LD_FINI_SWITCH
8305 If defined, a C string constant for a switch that tells the linker that
8306 the following symbol is a finalization routine.
8309 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8310 If defined, a C statement that will write a function that can be
8311 automatically called when a shared library is loaded. The function
8312 should call @var{func}, which takes no arguments. If not defined, and
8313 the object format requires an explicit initialization function, then a
8314 function called @code{_GLOBAL__DI} will be generated.
8316 This function and the following one are used by collect2 when linking a
8317 shared library that needs constructors or destructors, or has DWARF2
8318 exception tables embedded in the code.
8321 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8322 If defined, a C statement that will write a function that can be
8323 automatically called when a shared library is unloaded. The function
8324 should call @var{func}, which takes no arguments. If not defined, and
8325 the object format requires an explicit finalization function, then a
8326 function called @code{_GLOBAL__DD} will be generated.
8329 @defmac INVOKE__main
8330 If defined, @code{main} will call @code{__main} despite the presence of
8331 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8332 where the init section is not actually run automatically, but is still
8333 useful for collecting the lists of constructors and destructors.
8336 @defmac SUPPORTS_INIT_PRIORITY
8337 If nonzero, the C++ @code{init_priority} attribute is supported and the
8338 compiler should emit instructions to control the order of initialization
8339 of objects. If zero, the compiler will issue an error message upon
8340 encountering an @code{init_priority} attribute.
8343 @hook TARGET_HAVE_CTORS_DTORS
8344 This value is true if the target supports some ``native'' method of
8345 collecting constructors and destructors to be run at startup and exit.
8346 It is false if we must use @command{collect2}.
8349 @hook TARGET_ASM_CONSTRUCTOR
8350 If defined, a function that outputs assembler code to arrange to call
8351 the function referenced by @var{symbol} at initialization time.
8353 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8354 no arguments and with no return value. If the target supports initialization
8355 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8356 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8358 If this macro is not defined by the target, a suitable default will
8359 be chosen if (1) the target supports arbitrary section names, (2) the
8360 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8364 @hook TARGET_ASM_DESTRUCTOR
8365 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8366 functions rather than initialization functions.
8369 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8370 generated for the generated object file will have static linkage.
8372 If your system uses @command{collect2} as the means of processing
8373 constructors, then that program normally uses @command{nm} to scan
8374 an object file for constructor functions to be called.
8376 On certain kinds of systems, you can define this macro to make
8377 @command{collect2} work faster (and, in some cases, make it work at all):
8379 @defmac OBJECT_FORMAT_COFF
8380 Define this macro if the system uses COFF (Common Object File Format)
8381 object files, so that @command{collect2} can assume this format and scan
8382 object files directly for dynamic constructor/destructor functions.
8384 This macro is effective only in a native compiler; @command{collect2} as
8385 part of a cross compiler always uses @command{nm} for the target machine.
8388 @defmac REAL_NM_FILE_NAME
8389 Define this macro as a C string constant containing the file name to use
8390 to execute @command{nm}. The default is to search the path normally for
8393 If your system supports shared libraries and has a program to list the
8394 dynamic dependencies of a given library or executable, you can define
8395 these macros to enable support for running initialization and
8396 termination functions in shared libraries:
8400 Define this macro to a C string constant containing the name of the program
8401 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
8404 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8405 Define this macro to be C code that extracts filenames from the output
8406 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8407 of type @code{char *} that points to the beginning of a line of output
8408 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8409 code must advance @var{ptr} to the beginning of the filename on that
8410 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8413 @defmac SHLIB_SUFFIX
8414 Define this macro to a C string constant containing the default shared
8415 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8416 strips version information after this suffix when generating global
8417 constructor and destructor names. This define is only needed on targets
8418 that use @command{collect2} to process constructors and destructors.
8421 @node Instruction Output
8422 @subsection Output of Assembler Instructions
8424 @c prevent bad page break with this line
8425 This describes assembler instruction output.
8427 @defmac REGISTER_NAMES
8428 A C initializer containing the assembler's names for the machine
8429 registers, each one as a C string constant. This is what translates
8430 register numbers in the compiler into assembler language.
8433 @defmac ADDITIONAL_REGISTER_NAMES
8434 If defined, a C initializer for an array of structures containing a name
8435 and a register number. This macro defines additional names for hard
8436 registers, thus allowing the @code{asm} option in declarations to refer
8437 to registers using alternate names.
8440 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8441 Define this macro if you are using an unusual assembler that
8442 requires different names for the machine instructions.
8444 The definition is a C statement or statements which output an
8445 assembler instruction opcode to the stdio stream @var{stream}. The
8446 macro-operand @var{ptr} is a variable of type @code{char *} which
8447 points to the opcode name in its ``internal'' form---the form that is
8448 written in the machine description. The definition should output the
8449 opcode name to @var{stream}, performing any translation you desire, and
8450 increment the variable @var{ptr} to point at the end of the opcode
8451 so that it will not be output twice.
8453 In fact, your macro definition may process less than the entire opcode
8454 name, or more than the opcode name; but if you want to process text
8455 that includes @samp{%}-sequences to substitute operands, you must take
8456 care of the substitution yourself. Just be sure to increment
8457 @var{ptr} over whatever text should not be output normally.
8459 @findex recog_data.operand
8460 If you need to look at the operand values, they can be found as the
8461 elements of @code{recog_data.operand}.
8463 If the macro definition does nothing, the instruction is output
8467 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8468 If defined, a C statement to be executed just prior to the output of
8469 assembler code for @var{insn}, to modify the extracted operands so
8470 they will be output differently.
8472 Here the argument @var{opvec} is the vector containing the operands
8473 extracted from @var{insn}, and @var{noperands} is the number of
8474 elements of the vector which contain meaningful data for this insn.
8475 The contents of this vector are what will be used to convert the insn
8476 template into assembler code, so you can change the assembler output
8477 by changing the contents of the vector.
8479 This macro is useful when various assembler syntaxes share a single
8480 file of instruction patterns; by defining this macro differently, you
8481 can cause a large class of instructions to be output differently (such
8482 as with rearranged operands). Naturally, variations in assembler
8483 syntax affecting individual insn patterns ought to be handled by
8484 writing conditional output routines in those patterns.
8486 If this macro is not defined, it is equivalent to a null statement.
8489 @hook TARGET_ASM_FINAL_POSTSCAN_INSN
8490 If defined, this target hook is a function which is executed just after the
8491 output of assembler code for @var{insn}, to change the mode of the assembler
8494 Here the argument @var{opvec} is the vector containing the operands
8495 extracted from @var{insn}, and @var{noperands} is the number of
8496 elements of the vector which contain meaningful data for this insn.
8497 The contents of this vector are what was used to convert the insn
8498 template into assembler code, so you can change the assembler mode
8499 by checking the contents of the vector.
8502 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8503 A C compound statement to output to stdio stream @var{stream} the
8504 assembler syntax for an instruction operand @var{x}. @var{x} is an
8507 @var{code} is a value that can be used to specify one of several ways
8508 of printing the operand. It is used when identical operands must be
8509 printed differently depending on the context. @var{code} comes from
8510 the @samp{%} specification that was used to request printing of the
8511 operand. If the specification was just @samp{%@var{digit}} then
8512 @var{code} is 0; if the specification was @samp{%@var{ltr}
8513 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8516 If @var{x} is a register, this macro should print the register's name.
8517 The names can be found in an array @code{reg_names} whose type is
8518 @code{char *[]}. @code{reg_names} is initialized from
8519 @code{REGISTER_NAMES}.
8521 When the machine description has a specification @samp{%@var{punct}}
8522 (a @samp{%} followed by a punctuation character), this macro is called
8523 with a null pointer for @var{x} and the punctuation character for
8527 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8528 A C expression which evaluates to true if @var{code} is a valid
8529 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8530 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8531 punctuation characters (except for the standard one, @samp{%}) are used
8535 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8536 A C compound statement to output to stdio stream @var{stream} the
8537 assembler syntax for an instruction operand that is a memory reference
8538 whose address is @var{x}. @var{x} is an RTL expression.
8540 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8541 On some machines, the syntax for a symbolic address depends on the
8542 section that the address refers to. On these machines, define the hook
8543 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8544 @code{symbol_ref}, and then check for it here. @xref{Assembler
8548 @findex dbr_sequence_length
8549 @defmac DBR_OUTPUT_SEQEND (@var{file})
8550 A C statement, to be executed after all slot-filler instructions have
8551 been output. If necessary, call @code{dbr_sequence_length} to
8552 determine the number of slots filled in a sequence (zero if not
8553 currently outputting a sequence), to decide how many no-ops to output,
8556 Don't define this macro if it has nothing to do, but it is helpful in
8557 reading assembly output if the extent of the delay sequence is made
8558 explicit (e.g.@: with white space).
8561 @findex final_sequence
8562 Note that output routines for instructions with delay slots must be
8563 prepared to deal with not being output as part of a sequence
8564 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8565 found.) The variable @code{final_sequence} is null when not
8566 processing a sequence, otherwise it contains the @code{sequence} rtx
8570 @defmac REGISTER_PREFIX
8571 @defmacx LOCAL_LABEL_PREFIX
8572 @defmacx USER_LABEL_PREFIX
8573 @defmacx IMMEDIATE_PREFIX
8574 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8575 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8576 @file{final.c}). These are useful when a single @file{md} file must
8577 support multiple assembler formats. In that case, the various @file{tm.h}
8578 files can define these macros differently.
8581 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8582 If defined this macro should expand to a series of @code{case}
8583 statements which will be parsed inside the @code{switch} statement of
8584 the @code{asm_fprintf} function. This allows targets to define extra
8585 printf formats which may useful when generating their assembler
8586 statements. Note that uppercase letters are reserved for future
8587 generic extensions to asm_fprintf, and so are not available to target
8588 specific code. The output file is given by the parameter @var{file}.
8589 The varargs input pointer is @var{argptr} and the rest of the format
8590 string, starting the character after the one that is being switched
8591 upon, is pointed to by @var{format}.
8594 @defmac ASSEMBLER_DIALECT
8595 If your target supports multiple dialects of assembler language (such as
8596 different opcodes), define this macro as a C expression that gives the
8597 numeric index of the assembler language dialect to use, with zero as the
8600 If this macro is defined, you may use constructs of the form
8602 @samp{@{option0|option1|option2@dots{}@}}
8605 in the output templates of patterns (@pxref{Output Template}) or in the
8606 first argument of @code{asm_fprintf}. This construct outputs
8607 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8608 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8609 within these strings retain their usual meaning. If there are fewer
8610 alternatives within the braces than the value of
8611 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8613 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8614 @samp{@}} do not have any special meaning when used in templates or
8615 operands to @code{asm_fprintf}.
8617 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8618 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8619 the variations in assembler language syntax with that mechanism. Define
8620 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8621 if the syntax variant are larger and involve such things as different
8622 opcodes or operand order.
8625 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8626 A C expression to output to @var{stream} some assembler code
8627 which will push hard register number @var{regno} onto the stack.
8628 The code need not be optimal, since this macro is used only when
8632 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8633 A C expression to output to @var{stream} some assembler code
8634 which will pop hard register number @var{regno} off of the stack.
8635 The code need not be optimal, since this macro is used only when
8639 @node Dispatch Tables
8640 @subsection Output of Dispatch Tables
8642 @c prevent bad page break with this line
8643 This concerns dispatch tables.
8645 @cindex dispatch table
8646 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8647 A C statement to output to the stdio stream @var{stream} an assembler
8648 pseudo-instruction to generate a difference between two labels.
8649 @var{value} and @var{rel} are the numbers of two internal labels. The
8650 definitions of these labels are output using
8651 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8652 way here. For example,
8655 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8656 @var{value}, @var{rel})
8659 You must provide this macro on machines where the addresses in a
8660 dispatch table are relative to the table's own address. If defined, GCC
8661 will also use this macro on all machines when producing PIC@.
8662 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8663 mode and flags can be read.
8666 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8667 This macro should be provided on machines where the addresses
8668 in a dispatch table are absolute.
8670 The definition should be a C statement to output to the stdio stream
8671 @var{stream} an assembler pseudo-instruction to generate a reference to
8672 a label. @var{value} is the number of an internal label whose
8673 definition is output using @code{(*targetm.asm_out.internal_label)}.
8677 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8681 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8682 Define this if the label before a jump-table needs to be output
8683 specially. The first three arguments are the same as for
8684 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8685 jump-table which follows (a @code{jump_insn} containing an
8686 @code{addr_vec} or @code{addr_diff_vec}).
8688 This feature is used on system V to output a @code{swbeg} statement
8691 If this macro is not defined, these labels are output with
8692 @code{(*targetm.asm_out.internal_label)}.
8695 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8696 Define this if something special must be output at the end of a
8697 jump-table. The definition should be a C statement to be executed
8698 after the assembler code for the table is written. It should write
8699 the appropriate code to stdio stream @var{stream}. The argument
8700 @var{table} is the jump-table insn, and @var{num} is the label-number
8701 of the preceding label.
8703 If this macro is not defined, nothing special is output at the end of
8707 @hook TARGET_ASM_EMIT_UNWIND_LABEL
8708 This target hook emits a label at the beginning of each FDE@. It
8709 should be defined on targets where FDEs need special labels, and it
8710 should write the appropriate label, for the FDE associated with the
8711 function declaration @var{decl}, to the stdio stream @var{stream}.
8712 The third argument, @var{for_eh}, is a boolean: true if this is for an
8713 exception table. The fourth argument, @var{empty}, is a boolean:
8714 true if this is a placeholder label for an omitted FDE@.
8716 The default is that FDEs are not given nonlocal labels.
8719 @hook TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL
8720 This target hook emits a label at the beginning of the exception table.
8721 It should be defined on targets where it is desirable for the table
8722 to be broken up according to function.
8724 The default is that no label is emitted.
8727 @hook TARGET_ASM_UNWIND_EMIT
8728 This target hook emits assembly directives required to unwind the
8729 given instruction. This is only used when TARGET_UNWIND_INFO is set.
8732 @node Exception Region Output
8733 @subsection Assembler Commands for Exception Regions
8735 @c prevent bad page break with this line
8737 This describes commands marking the start and the end of an exception
8740 @defmac EH_FRAME_SECTION_NAME
8741 If defined, a C string constant for the name of the section containing
8742 exception handling frame unwind information. If not defined, GCC will
8743 provide a default definition if the target supports named sections.
8744 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8746 You should define this symbol if your target supports DWARF 2 frame
8747 unwind information and the default definition does not work.
8750 @defmac EH_FRAME_IN_DATA_SECTION
8751 If defined, DWARF 2 frame unwind information will be placed in the
8752 data section even though the target supports named sections. This
8753 might be necessary, for instance, if the system linker does garbage
8754 collection and sections cannot be marked as not to be collected.
8756 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8760 @defmac EH_TABLES_CAN_BE_READ_ONLY
8761 Define this macro to 1 if your target is such that no frame unwind
8762 information encoding used with non-PIC code will ever require a
8763 runtime relocation, but the linker may not support merging read-only
8764 and read-write sections into a single read-write section.
8767 @defmac MASK_RETURN_ADDR
8768 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8769 that it does not contain any extraneous set bits in it.
8772 @defmac DWARF2_UNWIND_INFO
8773 Define this macro to 0 if your target supports DWARF 2 frame unwind
8774 information, but it does not yet work with exception handling.
8775 Otherwise, if your target supports this information (if it defines
8776 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
8777 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of 1.
8779 If @code{TARGET_UNWIND_INFO} is defined, the target specific unwinder
8780 will be used in all cases. Defining this macro will enable the generation
8781 of DWARF 2 frame debugging information.
8783 If @code{TARGET_UNWIND_INFO} is not defined, and this macro is defined to 1,
8784 the DWARF 2 unwinder will be the default exception handling mechanism;
8785 otherwise, the @code{setjmp}/@code{longjmp}-based scheme will be used by
8789 @defmac TARGET_UNWIND_INFO
8790 Define this macro if your target has ABI specified unwind tables. Usually
8791 these will be output by @code{TARGET_ASM_UNWIND_EMIT}.
8794 @hook TARGET_UNWIND_TABLES_DEFAULT
8795 This variable should be set to @code{true} if the target ABI requires unwinding
8796 tables even when exceptions are not used.
8799 @defmac MUST_USE_SJLJ_EXCEPTIONS
8800 This macro need only be defined if @code{DWARF2_UNWIND_INFO} is
8801 runtime-variable. In that case, @file{except.h} cannot correctly
8802 determine the corresponding definition of @code{MUST_USE_SJLJ_EXCEPTIONS},
8803 so the target must provide it directly.
8806 @defmac DONT_USE_BUILTIN_SETJMP
8807 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8808 should use the @code{setjmp}/@code{longjmp} functions from the C library
8809 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8812 @defmac DWARF_CIE_DATA_ALIGNMENT
8813 This macro need only be defined if the target might save registers in the
8814 function prologue at an offset to the stack pointer that is not aligned to
8815 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8816 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8817 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8818 the target supports DWARF 2 frame unwind information.
8821 @hook TARGET_TERMINATE_DW2_EH_FRAME_INFO
8822 Contains the value true if the target should add a zero word onto the
8823 end of a Dwarf-2 frame info section when used for exception handling.
8824 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8828 @hook TARGET_DWARF_REGISTER_SPAN
8829 Given a register, this hook should return a parallel of registers to
8830 represent where to find the register pieces. Define this hook if the
8831 register and its mode are represented in Dwarf in non-contiguous
8832 locations, or if the register should be represented in more than one
8833 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8834 If not defined, the default is to return @code{NULL_RTX}.
8837 @hook TARGET_INIT_DWARF_REG_SIZES_EXTRA
8838 If some registers are represented in Dwarf-2 unwind information in
8839 multiple pieces, define this hook to fill in information about the
8840 sizes of those pieces in the table used by the unwinder at runtime.
8841 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8842 filling in a single size corresponding to each hard register;
8843 @var{address} is the address of the table.
8846 @hook TARGET_ASM_TTYPE
8847 This hook is used to output a reference from a frame unwinding table to
8848 the type_info object identified by @var{sym}. It should return @code{true}
8849 if the reference was output. Returning @code{false} will cause the
8850 reference to be output using the normal Dwarf2 routines.
8853 @hook TARGET_ARM_EABI_UNWINDER
8854 This flag should be set to @code{true} on targets that use an ARM EABI
8855 based unwinding library, and @code{false} on other targets. This effects
8856 the format of unwinding tables, and how the unwinder in entered after
8857 running a cleanup. The default is @code{false}.
8860 @node Alignment Output
8861 @subsection Assembler Commands for Alignment
8863 @c prevent bad page break with this line
8864 This describes commands for alignment.
8866 @defmac JUMP_ALIGN (@var{label})
8867 The alignment (log base 2) to put in front of @var{label}, which is
8868 a common destination of jumps and has no fallthru incoming edge.
8870 This macro need not be defined if you don't want any special alignment
8871 to be done at such a time. Most machine descriptions do not currently
8874 Unless it's necessary to inspect the @var{label} parameter, it is better
8875 to set the variable @var{align_jumps} in the target's
8876 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8877 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8880 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8881 The alignment (log base 2) to put in front of @var{label}, which follows
8884 This macro need not be defined if you don't want any special alignment
8885 to be done at such a time. Most machine descriptions do not currently
8889 @defmac LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
8890 The maximum number of bytes to skip when applying
8891 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8892 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8895 @defmac LOOP_ALIGN (@var{label})
8896 The alignment (log base 2) to put in front of @var{label}, which follows
8897 a @code{NOTE_INSN_LOOP_BEG} note.
8899 This macro need not be defined if you don't want any special alignment
8900 to be done at such a time. Most machine descriptions do not currently
8903 Unless it's necessary to inspect the @var{label} parameter, it is better
8904 to set the variable @code{align_loops} in the target's
8905 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8906 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
8909 @defmac LOOP_ALIGN_MAX_SKIP
8910 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
8911 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8914 @defmac LABEL_ALIGN (@var{label})
8915 The alignment (log base 2) to put in front of @var{label}.
8916 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
8917 the maximum of the specified values is used.
8919 Unless it's necessary to inspect the @var{label} parameter, it is better
8920 to set the variable @code{align_labels} in the target's
8921 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8922 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
8925 @defmac LABEL_ALIGN_MAX_SKIP
8926 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
8927 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8930 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
8931 A C statement to output to the stdio stream @var{stream} an assembler
8932 instruction to advance the location counter by @var{nbytes} bytes.
8933 Those bytes should be zero when loaded. @var{nbytes} will be a C
8934 expression of type @code{unsigned HOST_WIDE_INT}.
8937 @defmac ASM_NO_SKIP_IN_TEXT
8938 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
8939 text section because it fails to put zeros in the bytes that are skipped.
8940 This is true on many Unix systems, where the pseudo--op to skip bytes
8941 produces no-op instructions rather than zeros when used in the text
8945 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
8946 A C statement to output to the stdio stream @var{stream} an assembler
8947 command to advance the location counter to a multiple of 2 to the
8948 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
8951 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
8952 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
8953 for padding, if necessary.
8956 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
8957 A C statement to output to the stdio stream @var{stream} an assembler
8958 command to advance the location counter to a multiple of 2 to the
8959 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
8960 satisfy the alignment request. @var{power} and @var{max_skip} will be
8961 a C expression of type @code{int}.
8965 @node Debugging Info
8966 @section Controlling Debugging Information Format
8968 @c prevent bad page break with this line
8969 This describes how to specify debugging information.
8972 * All Debuggers:: Macros that affect all debugging formats uniformly.
8973 * DBX Options:: Macros enabling specific options in DBX format.
8974 * DBX Hooks:: Hook macros for varying DBX format.
8975 * File Names and DBX:: Macros controlling output of file names in DBX format.
8976 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
8977 * VMS Debug:: Macros for VMS debug format.
8981 @subsection Macros Affecting All Debugging Formats
8983 @c prevent bad page break with this line
8984 These macros affect all debugging formats.
8986 @defmac DBX_REGISTER_NUMBER (@var{regno})
8987 A C expression that returns the DBX register number for the compiler
8988 register number @var{regno}. In the default macro provided, the value
8989 of this expression will be @var{regno} itself. But sometimes there are
8990 some registers that the compiler knows about and DBX does not, or vice
8991 versa. In such cases, some register may need to have one number in the
8992 compiler and another for DBX@.
8994 If two registers have consecutive numbers inside GCC, and they can be
8995 used as a pair to hold a multiword value, then they @emph{must} have
8996 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
8997 Otherwise, debuggers will be unable to access such a pair, because they
8998 expect register pairs to be consecutive in their own numbering scheme.
9000 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
9001 does not preserve register pairs, then what you must do instead is
9002 redefine the actual register numbering scheme.
9005 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
9006 A C expression that returns the integer offset value for an automatic
9007 variable having address @var{x} (an RTL expression). The default
9008 computation assumes that @var{x} is based on the frame-pointer and
9009 gives the offset from the frame-pointer. This is required for targets
9010 that produce debugging output for DBX or COFF-style debugging output
9011 for SDB and allow the frame-pointer to be eliminated when the
9012 @option{-g} options is used.
9015 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
9016 A C expression that returns the integer offset value for an argument
9017 having address @var{x} (an RTL expression). The nominal offset is
9021 @defmac PREFERRED_DEBUGGING_TYPE
9022 A C expression that returns the type of debugging output GCC should
9023 produce when the user specifies just @option{-g}. Define
9024 this if you have arranged for GCC to support more than one format of
9025 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
9026 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
9027 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
9029 When the user specifies @option{-ggdb}, GCC normally also uses the
9030 value of this macro to select the debugging output format, but with two
9031 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
9032 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
9033 defined, GCC uses @code{DBX_DEBUG}.
9035 The value of this macro only affects the default debugging output; the
9036 user can always get a specific type of output by using @option{-gstabs},
9037 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
9041 @subsection Specific Options for DBX Output
9043 @c prevent bad page break with this line
9044 These are specific options for DBX output.
9046 @defmac DBX_DEBUGGING_INFO
9047 Define this macro if GCC should produce debugging output for DBX
9048 in response to the @option{-g} option.
9051 @defmac XCOFF_DEBUGGING_INFO
9052 Define this macro if GCC should produce XCOFF format debugging output
9053 in response to the @option{-g} option. This is a variant of DBX format.
9056 @defmac DEFAULT_GDB_EXTENSIONS
9057 Define this macro to control whether GCC should by default generate
9058 GDB's extended version of DBX debugging information (assuming DBX-format
9059 debugging information is enabled at all). If you don't define the
9060 macro, the default is 1: always generate the extended information
9061 if there is any occasion to.
9064 @defmac DEBUG_SYMS_TEXT
9065 Define this macro if all @code{.stabs} commands should be output while
9066 in the text section.
9069 @defmac ASM_STABS_OP
9070 A C string constant, including spacing, naming the assembler pseudo op to
9071 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
9072 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
9073 applies only to DBX debugging information format.
9076 @defmac ASM_STABD_OP
9077 A C string constant, including spacing, naming the assembler pseudo op to
9078 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
9079 value is the current location. If you don't define this macro,
9080 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
9084 @defmac ASM_STABN_OP
9085 A C string constant, including spacing, naming the assembler pseudo op to
9086 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
9087 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
9088 macro applies only to DBX debugging information format.
9091 @defmac DBX_NO_XREFS
9092 Define this macro if DBX on your system does not support the construct
9093 @samp{xs@var{tagname}}. On some systems, this construct is used to
9094 describe a forward reference to a structure named @var{tagname}.
9095 On other systems, this construct is not supported at all.
9098 @defmac DBX_CONTIN_LENGTH
9099 A symbol name in DBX-format debugging information is normally
9100 continued (split into two separate @code{.stabs} directives) when it
9101 exceeds a certain length (by default, 80 characters). On some
9102 operating systems, DBX requires this splitting; on others, splitting
9103 must not be done. You can inhibit splitting by defining this macro
9104 with the value zero. You can override the default splitting-length by
9105 defining this macro as an expression for the length you desire.
9108 @defmac DBX_CONTIN_CHAR
9109 Normally continuation is indicated by adding a @samp{\} character to
9110 the end of a @code{.stabs} string when a continuation follows. To use
9111 a different character instead, define this macro as a character
9112 constant for the character you want to use. Do not define this macro
9113 if backslash is correct for your system.
9116 @defmac DBX_STATIC_STAB_DATA_SECTION
9117 Define this macro if it is necessary to go to the data section before
9118 outputting the @samp{.stabs} pseudo-op for a non-global static
9122 @defmac DBX_TYPE_DECL_STABS_CODE
9123 The value to use in the ``code'' field of the @code{.stabs} directive
9124 for a typedef. The default is @code{N_LSYM}.
9127 @defmac DBX_STATIC_CONST_VAR_CODE
9128 The value to use in the ``code'' field of the @code{.stabs} directive
9129 for a static variable located in the text section. DBX format does not
9130 provide any ``right'' way to do this. The default is @code{N_FUN}.
9133 @defmac DBX_REGPARM_STABS_CODE
9134 The value to use in the ``code'' field of the @code{.stabs} directive
9135 for a parameter passed in registers. DBX format does not provide any
9136 ``right'' way to do this. The default is @code{N_RSYM}.
9139 @defmac DBX_REGPARM_STABS_LETTER
9140 The letter to use in DBX symbol data to identify a symbol as a parameter
9141 passed in registers. DBX format does not customarily provide any way to
9142 do this. The default is @code{'P'}.
9145 @defmac DBX_FUNCTION_FIRST
9146 Define this macro if the DBX information for a function and its
9147 arguments should precede the assembler code for the function. Normally,
9148 in DBX format, the debugging information entirely follows the assembler
9152 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
9153 Define this macro, with value 1, if the value of a symbol describing
9154 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
9155 relative to the start of the enclosing function. Normally, GCC uses
9156 an absolute address.
9159 @defmac DBX_LINES_FUNCTION_RELATIVE
9160 Define this macro, with value 1, if the value of a symbol indicating
9161 the current line number (@code{N_SLINE}) should be relative to the
9162 start of the enclosing function. Normally, GCC uses an absolute address.
9165 @defmac DBX_USE_BINCL
9166 Define this macro if GCC should generate @code{N_BINCL} and
9167 @code{N_EINCL} stabs for included header files, as on Sun systems. This
9168 macro also directs GCC to output a type number as a pair of a file
9169 number and a type number within the file. Normally, GCC does not
9170 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9171 number for a type number.
9175 @subsection Open-Ended Hooks for DBX Format
9177 @c prevent bad page break with this line
9178 These are hooks for DBX format.
9180 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
9181 Define this macro to say how to output to @var{stream} the debugging
9182 information for the start of a scope level for variable names. The
9183 argument @var{name} is the name of an assembler symbol (for use with
9184 @code{assemble_name}) whose value is the address where the scope begins.
9187 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
9188 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
9191 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
9192 Define this macro if the target machine requires special handling to
9193 output an @code{N_FUN} entry for the function @var{decl}.
9196 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9197 A C statement to output DBX debugging information before code for line
9198 number @var{line} of the current source file to the stdio stream
9199 @var{stream}. @var{counter} is the number of time the macro was
9200 invoked, including the current invocation; it is intended to generate
9201 unique labels in the assembly output.
9203 This macro should not be defined if the default output is correct, or
9204 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9207 @defmac NO_DBX_FUNCTION_END
9208 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9209 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9210 On those machines, define this macro to turn this feature off without
9211 disturbing the rest of the gdb extensions.
9214 @defmac NO_DBX_BNSYM_ENSYM
9215 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9216 extension construct. On those machines, define this macro to turn this
9217 feature off without disturbing the rest of the gdb extensions.
9220 @node File Names and DBX
9221 @subsection File Names in DBX Format
9223 @c prevent bad page break with this line
9224 This describes file names in DBX format.
9226 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9227 A C statement to output DBX debugging information to the stdio stream
9228 @var{stream}, which indicates that file @var{name} is the main source
9229 file---the file specified as the input file for compilation.
9230 This macro is called only once, at the beginning of compilation.
9232 This macro need not be defined if the standard form of output
9233 for DBX debugging information is appropriate.
9235 It may be necessary to refer to a label equal to the beginning of the
9236 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9237 to do so. If you do this, you must also set the variable
9238 @var{used_ltext_label_name} to @code{true}.
9241 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9242 Define this macro, with value 1, if GCC should not emit an indication
9243 of the current directory for compilation and current source language at
9244 the beginning of the file.
9247 @defmac NO_DBX_GCC_MARKER
9248 Define this macro, with value 1, if GCC should not emit an indication
9249 that this object file was compiled by GCC@. The default is to emit
9250 an @code{N_OPT} stab at the beginning of every source file, with
9251 @samp{gcc2_compiled.} for the string and value 0.
9254 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9255 A C statement to output DBX debugging information at the end of
9256 compilation of the main source file @var{name}. Output should be
9257 written to the stdio stream @var{stream}.
9259 If you don't define this macro, nothing special is output at the end
9260 of compilation, which is correct for most machines.
9263 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9264 Define this macro @emph{instead of} defining
9265 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9266 the end of compilation is an @code{N_SO} stab with an empty string,
9267 whose value is the highest absolute text address in the file.
9272 @subsection Macros for SDB and DWARF Output
9274 @c prevent bad page break with this line
9275 Here are macros for SDB and DWARF output.
9277 @defmac SDB_DEBUGGING_INFO
9278 Define this macro if GCC should produce COFF-style debugging output
9279 for SDB in response to the @option{-g} option.
9282 @defmac DWARF2_DEBUGGING_INFO
9283 Define this macro if GCC should produce dwarf version 2 format
9284 debugging output in response to the @option{-g} option.
9286 @hook TARGET_DWARF_CALLING_CONVENTION
9287 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9288 be emitted for each function. Instead of an integer return the enum
9289 value for the @code{DW_CC_} tag.
9292 To support optional call frame debugging information, you must also
9293 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9294 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9295 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9296 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9299 @defmac DWARF2_FRAME_INFO
9300 Define this macro to a nonzero value if GCC should always output
9301 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
9302 (@pxref{Exception Region Output} is nonzero, GCC will output this
9303 information not matter how you define @code{DWARF2_FRAME_INFO}.
9306 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9307 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9308 line debug info sections. This will result in much more compact line number
9309 tables, and hence is desirable if it works.
9312 @hook TARGET_WANT_DEBUG_PUB_SECTIONS
9314 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9315 A C statement to issue assembly directives that create a difference
9316 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9319 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9320 A C statement to issue assembly directives that create a difference
9321 between the two given labels in system defined units, e.g. instruction
9322 slots on IA64 VMS, using an integer of the given size.
9325 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9326 A C statement to issue assembly directives that create a
9327 section-relative reference to the given @var{label}, using an integer of the
9328 given @var{size}. The label is known to be defined in the given @var{section}.
9331 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9332 A C statement to issue assembly directives that create a self-relative
9333 reference to the given @var{label}, using an integer of the given @var{size}.
9336 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
9337 A C statement to issue assembly directives that create a reference to
9338 the DWARF table identifier @var{label} from the current section. This
9339 is used on some systems to avoid garbage collecting a DWARF table which
9340 is referenced by a function.
9343 @hook TARGET_ASM_OUTPUT_DWARF_DTPREL
9344 If defined, this target hook is a function which outputs a DTP-relative
9345 reference to the given TLS symbol of the specified size.
9348 @defmac PUT_SDB_@dots{}
9349 Define these macros to override the assembler syntax for the special
9350 SDB assembler directives. See @file{sdbout.c} for a list of these
9351 macros and their arguments. If the standard syntax is used, you need
9352 not define them yourself.
9356 Some assemblers do not support a semicolon as a delimiter, even between
9357 SDB assembler directives. In that case, define this macro to be the
9358 delimiter to use (usually @samp{\n}). It is not necessary to define
9359 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9363 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9364 Define this macro to allow references to unknown structure,
9365 union, or enumeration tags to be emitted. Standard COFF does not
9366 allow handling of unknown references, MIPS ECOFF has support for
9370 @defmac SDB_ALLOW_FORWARD_REFERENCES
9371 Define this macro to allow references to structure, union, or
9372 enumeration tags that have not yet been seen to be handled. Some
9373 assemblers choke if forward tags are used, while some require it.
9376 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9377 A C statement to output SDB debugging information before code for line
9378 number @var{line} of the current source file to the stdio stream
9379 @var{stream}. The default is to emit an @code{.ln} directive.
9384 @subsection Macros for VMS Debug Format
9386 @c prevent bad page break with this line
9387 Here are macros for VMS debug format.
9389 @defmac VMS_DEBUGGING_INFO
9390 Define this macro if GCC should produce debugging output for VMS
9391 in response to the @option{-g} option. The default behavior for VMS
9392 is to generate minimal debug info for a traceback in the absence of
9393 @option{-g} unless explicitly overridden with @option{-g0}. This
9394 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
9395 @code{TARGET_OPTION_OVERRIDE}.
9398 @node Floating Point
9399 @section Cross Compilation and Floating Point
9400 @cindex cross compilation and floating point
9401 @cindex floating point and cross compilation
9403 While all modern machines use twos-complement representation for integers,
9404 there are a variety of representations for floating point numbers. This
9405 means that in a cross-compiler the representation of floating point numbers
9406 in the compiled program may be different from that used in the machine
9407 doing the compilation.
9409 Because different representation systems may offer different amounts of
9410 range and precision, all floating point constants must be represented in
9411 the target machine's format. Therefore, the cross compiler cannot
9412 safely use the host machine's floating point arithmetic; it must emulate
9413 the target's arithmetic. To ensure consistency, GCC always uses
9414 emulation to work with floating point values, even when the host and
9415 target floating point formats are identical.
9417 The following macros are provided by @file{real.h} for the compiler to
9418 use. All parts of the compiler which generate or optimize
9419 floating-point calculations must use these macros. They may evaluate
9420 their operands more than once, so operands must not have side effects.
9422 @defmac REAL_VALUE_TYPE
9423 The C data type to be used to hold a floating point value in the target
9424 machine's format. Typically this is a @code{struct} containing an
9425 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9429 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9430 Compares for equality the two values, @var{x} and @var{y}. If the target
9431 floating point format supports negative zeroes and/or NaNs,
9432 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9433 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9436 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9437 Tests whether @var{x} is less than @var{y}.
9440 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9441 Truncates @var{x} to a signed integer, rounding toward zero.
9444 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9445 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9446 @var{x} is negative, returns zero.
9449 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9450 Converts @var{string} into a floating point number in the target machine's
9451 representation for mode @var{mode}. This routine can handle both
9452 decimal and hexadecimal floating point constants, using the syntax
9453 defined by the C language for both.
9456 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9457 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9460 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9461 Determines whether @var{x} represents infinity (positive or negative).
9464 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9465 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9468 @deftypefn Macro void REAL_ARITHMETIC (REAL_VALUE_TYPE @var{output}, enum tree_code @var{code}, REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9469 Calculates an arithmetic operation on the two floating point values
9470 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9473 The operation to be performed is specified by @var{code}. Only the
9474 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9475 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9477 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9478 target's floating point format cannot represent infinity, it will call
9479 @code{abort}. Callers should check for this situation first, using
9480 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9483 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9484 Returns the negative of the floating point value @var{x}.
9487 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9488 Returns the absolute value of @var{x}.
9491 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
9492 Truncates the floating point value @var{x} to fit in @var{mode}. The
9493 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
9494 appropriate bit pattern to be output as a floating constant whose
9495 precision accords with mode @var{mode}.
9498 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9499 Converts a floating point value @var{x} into a double-precision integer
9500 which is then stored into @var{low} and @var{high}. If the value is not
9501 integral, it is truncated.
9504 @deftypefn Macro void REAL_VALUE_FROM_INT (REAL_VALUE_TYPE @var{x}, HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, enum machine_mode @var{mode})
9505 Converts a double-precision integer found in @var{low} and @var{high},
9506 into a floating point value which is then stored into @var{x}. The
9507 value is truncated to fit in mode @var{mode}.
9510 @node Mode Switching
9511 @section Mode Switching Instructions
9512 @cindex mode switching
9513 The following macros control mode switching optimizations:
9515 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9516 Define this macro if the port needs extra instructions inserted for mode
9517 switching in an optimizing compilation.
9519 For an example, the SH4 can perform both single and double precision
9520 floating point operations, but to perform a single precision operation,
9521 the FPSCR PR bit has to be cleared, while for a double precision
9522 operation, this bit has to be set. Changing the PR bit requires a general
9523 purpose register as a scratch register, hence these FPSCR sets have to
9524 be inserted before reload, i.e.@: you can't put this into instruction emitting
9525 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9527 You can have multiple entities that are mode-switched, and select at run time
9528 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9529 return nonzero for any @var{entity} that needs mode-switching.
9530 If you define this macro, you also have to define
9531 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9532 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9533 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9537 @defmac NUM_MODES_FOR_MODE_SWITCHING
9538 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9539 initializer for an array of integers. Each initializer element
9540 N refers to an entity that needs mode switching, and specifies the number
9541 of different modes that might need to be set for this entity.
9542 The position of the initializer in the initializer---starting counting at
9543 zero---determines the integer that is used to refer to the mode-switched
9545 In macros that take mode arguments / yield a mode result, modes are
9546 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9547 switch is needed / supplied.
9550 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9551 @var{entity} is an integer specifying a mode-switched entity. If
9552 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9553 return an integer value not larger than the corresponding element in
9554 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9555 be switched into prior to the execution of @var{insn}.
9558 @defmac MODE_AFTER (@var{mode}, @var{insn})
9559 If this macro is defined, it is evaluated for every @var{insn} during
9560 mode switching. It determines the mode that an insn results in (if
9561 different from the incoming mode).
9564 @defmac MODE_ENTRY (@var{entity})
9565 If this macro is defined, it is evaluated for every @var{entity} that needs
9566 mode switching. It should evaluate to an integer, which is a mode that
9567 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9568 is defined then @code{MODE_EXIT} must be defined.
9571 @defmac MODE_EXIT (@var{entity})
9572 If this macro is defined, it is evaluated for every @var{entity} that needs
9573 mode switching. It should evaluate to an integer, which is a mode that
9574 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9575 is defined then @code{MODE_ENTRY} must be defined.
9578 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9579 This macro specifies the order in which modes for @var{entity} are processed.
9580 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9581 lowest. The value of the macro should be an integer designating a mode
9582 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9583 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9584 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9587 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9588 Generate one or more insns to set @var{entity} to @var{mode}.
9589 @var{hard_reg_live} is the set of hard registers live at the point where
9590 the insn(s) are to be inserted.
9593 @node Target Attributes
9594 @section Defining target-specific uses of @code{__attribute__}
9595 @cindex target attributes
9596 @cindex machine attributes
9597 @cindex attributes, target-specific
9599 Target-specific attributes may be defined for functions, data and types.
9600 These are described using the following target hooks; they also need to
9601 be documented in @file{extend.texi}.
9603 @hook TARGET_ATTRIBUTE_TABLE
9604 If defined, this target hook points to an array of @samp{struct
9605 attribute_spec} (defined in @file{tree.h}) specifying the machine
9606 specific attributes for this target and some of the restrictions on the
9607 entities to which these attributes are applied and the arguments they
9611 @hook TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P
9612 If defined, this target hook is a function which returns true if the
9613 machine-specific attribute named @var{name} expects an identifier
9614 given as its first argument to be passed on as a plain identifier, not
9615 subjected to name lookup. If this is not defined, the default is
9616 false for all machine-specific attributes.
9619 @hook TARGET_COMP_TYPE_ATTRIBUTES
9620 If defined, this target hook is a function which returns zero if the attributes on
9621 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9622 and two if they are nearly compatible (which causes a warning to be
9623 generated). If this is not defined, machine-specific attributes are
9624 supposed always to be compatible.
9627 @hook TARGET_SET_DEFAULT_TYPE_ATTRIBUTES
9628 If defined, this target hook is a function which assigns default attributes to
9629 the newly defined @var{type}.
9632 @hook TARGET_MERGE_TYPE_ATTRIBUTES
9633 Define this target hook if the merging of type attributes needs special
9634 handling. If defined, the result is a list of the combined
9635 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9636 that @code{comptypes} has already been called and returned 1. This
9637 function may call @code{merge_attributes} to handle machine-independent
9641 @hook TARGET_MERGE_DECL_ATTRIBUTES
9642 Define this target hook if the merging of decl attributes needs special
9643 handling. If defined, the result is a list of the combined
9644 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9645 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9646 when this is needed are when one attribute overrides another, or when an
9647 attribute is nullified by a subsequent definition. This function may
9648 call @code{merge_attributes} to handle machine-independent merging.
9650 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9651 If the only target-specific handling you require is @samp{dllimport}
9652 for Microsoft Windows targets, you should define the macro
9653 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9654 will then define a function called
9655 @code{merge_dllimport_decl_attributes} which can then be defined as
9656 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9657 add @code{handle_dll_attribute} in the attribute table for your port
9658 to perform initial processing of the @samp{dllimport} and
9659 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9660 @file{i386/i386.c}, for example.
9663 @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P
9665 @defmac TARGET_DECLSPEC
9666 Define this macro to a nonzero value if you want to treat
9667 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9668 default, this behavior is enabled only for targets that define
9669 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9670 of @code{__declspec} is via a built-in macro, but you should not rely
9671 on this implementation detail.
9674 @hook TARGET_INSERT_ATTRIBUTES
9675 Define this target hook if you want to be able to add attributes to a decl
9676 when it is being created. This is normally useful for back ends which
9677 wish to implement a pragma by using the attributes which correspond to
9678 the pragma's effect. The @var{node} argument is the decl which is being
9679 created. The @var{attr_ptr} argument is a pointer to the attribute list
9680 for this decl. The list itself should not be modified, since it may be
9681 shared with other decls, but attributes may be chained on the head of
9682 the list and @code{*@var{attr_ptr}} modified to point to the new
9683 attributes, or a copy of the list may be made if further changes are
9687 @hook TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P
9689 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9690 into the current function, despite its having target-specific
9691 attributes, @code{false} otherwise. By default, if a function has a
9692 target specific attribute attached to it, it will not be inlined.
9695 @hook TARGET_OPTION_VALID_ATTRIBUTE_P
9696 This hook is called to parse the @code{attribute(option("..."))}, and
9697 it allows the function to set different target machine compile time
9698 options for the current function that might be different than the
9699 options specified on the command line. The hook should return
9700 @code{true} if the options are valid.
9702 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9703 the function declaration to hold a pointer to a target specific
9704 @var{struct cl_target_option} structure.
9707 @hook TARGET_OPTION_SAVE
9708 This hook is called to save any additional target specific information
9709 in the @var{struct cl_target_option} structure for function specific
9711 @xref{Option file format}.
9714 @hook TARGET_OPTION_RESTORE
9715 This hook is called to restore any additional target specific
9716 information in the @var{struct cl_target_option} structure for
9717 function specific options.
9720 @hook TARGET_OPTION_PRINT
9721 This hook is called to print any additional target specific
9722 information in the @var{struct cl_target_option} structure for
9723 function specific options.
9726 @hook TARGET_OPTION_PRAGMA_PARSE
9727 This target hook parses the options for @code{#pragma GCC option} to
9728 set the machine specific options for functions that occur later in the
9729 input stream. The options should be the same as handled by the
9730 @code{TARGET_OPTION_VALID_ATTRIBUTE_P} hook.
9733 @hook TARGET_OPTION_OVERRIDE
9734 Sometimes certain combinations of command options do not make sense on
9735 a particular target machine. You can override the hook
9736 @code{TARGET_OPTION_OVERRIDE} to take account of this. This hooks is called
9737 once just after all the command options have been parsed.
9739 Don't use this hook to turn on various extra optimizations for
9740 @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
9742 If you need to do something whenever the optimization level is
9743 changed via the optimize attribute or pragma, see
9744 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}
9747 @hook TARGET_CAN_INLINE_P
9748 This target hook returns @code{false} if the @var{caller} function
9749 cannot inline @var{callee}, based on target specific information. By
9750 default, inlining is not allowed if the callee function has function
9751 specific target options and the caller does not use the same options.
9755 @section Emulating TLS
9756 @cindex Emulated TLS
9758 For targets whose psABI does not provide Thread Local Storage via
9759 specific relocations and instruction sequences, an emulation layer is
9760 used. A set of target hooks allows this emulation layer to be
9761 configured for the requirements of a particular target. For instance
9762 the psABI may in fact specify TLS support in terms of an emulation
9765 The emulation layer works by creating a control object for every TLS
9766 object. To access the TLS object, a lookup function is provided
9767 which, when given the address of the control object, will return the
9768 address of the current thread's instance of the TLS object.
9770 @hook TARGET_EMUTLS_GET_ADDRESS
9771 Contains the name of the helper function that uses a TLS control
9772 object to locate a TLS instance. The default causes libgcc's
9773 emulated TLS helper function to be used.
9776 @hook TARGET_EMUTLS_REGISTER_COMMON
9777 Contains the name of the helper function that should be used at
9778 program startup to register TLS objects that are implicitly
9779 initialized to zero. If this is @code{NULL}, all TLS objects will
9780 have explicit initializers. The default causes libgcc's emulated TLS
9781 registration function to be used.
9784 @hook TARGET_EMUTLS_VAR_SECTION
9785 Contains the name of the section in which TLS control variables should
9786 be placed. The default of @code{NULL} allows these to be placed in
9790 @hook TARGET_EMUTLS_TMPL_SECTION
9791 Contains the name of the section in which TLS initializers should be
9792 placed. The default of @code{NULL} allows these to be placed in any
9796 @hook TARGET_EMUTLS_VAR_PREFIX
9797 Contains the prefix to be prepended to TLS control variable names.
9798 The default of @code{NULL} uses a target-specific prefix.
9801 @hook TARGET_EMUTLS_TMPL_PREFIX
9802 Contains the prefix to be prepended to TLS initializer objects. The
9803 default of @code{NULL} uses a target-specific prefix.
9806 @hook TARGET_EMUTLS_VAR_FIELDS
9807 Specifies a function that generates the FIELD_DECLs for a TLS control
9808 object type. @var{type} is the RECORD_TYPE the fields are for and
9809 @var{name} should be filled with the structure tag, if the default of
9810 @code{__emutls_object} is unsuitable. The default creates a type suitable
9811 for libgcc's emulated TLS function.
9814 @hook TARGET_EMUTLS_VAR_INIT
9815 Specifies a function that generates the CONSTRUCTOR to initialize a
9816 TLS control object. @var{var} is the TLS control object, @var{decl}
9817 is the TLS object and @var{tmpl_addr} is the address of the
9818 initializer. The default initializes libgcc's emulated TLS control object.
9821 @hook TARGET_EMUTLS_VAR_ALIGN_FIXED
9822 Specifies whether the alignment of TLS control variable objects is
9823 fixed and should not be increased as some backends may do to optimize
9824 single objects. The default is false.
9827 @hook TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9828 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9829 may be used to describe emulated TLS control objects.
9832 @node MIPS Coprocessors
9833 @section Defining coprocessor specifics for MIPS targets.
9834 @cindex MIPS coprocessor-definition macros
9836 The MIPS specification allows MIPS implementations to have as many as 4
9837 coprocessors, each with as many as 32 private registers. GCC supports
9838 accessing these registers and transferring values between the registers
9839 and memory using asm-ized variables. For example:
9842 register unsigned int cp0count asm ("c0r1");
9848 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9849 names may be added as described below, or the default names may be
9850 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9852 Coprocessor registers are assumed to be epilogue-used; sets to them will
9853 be preserved even if it does not appear that the register is used again
9854 later in the function.
9856 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9857 the FPU@. One accesses COP1 registers through standard mips
9858 floating-point support; they are not included in this mechanism.
9860 There is one macro used in defining the MIPS coprocessor interface which
9861 you may want to override in subtargets; it is described below.
9863 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
9864 A comma-separated list (with leading comma) of pairs describing the
9865 alternate names of coprocessor registers. The format of each entry should be
9867 @{ @var{alternatename}, @var{register_number}@}
9873 @section Parameters for Precompiled Header Validity Checking
9874 @cindex parameters, precompiled headers
9876 @hook TARGET_GET_PCH_VALIDITY
9877 This hook returns a pointer to the data needed by
9878 @code{TARGET_PCH_VALID_P} and sets
9879 @samp{*@var{sz}} to the size of the data in bytes.
9882 @hook TARGET_PCH_VALID_P
9883 This hook checks whether the options used to create a PCH file are
9884 compatible with the current settings. It returns @code{NULL}
9885 if so and a suitable error message if not. Error messages will
9886 be presented to the user and must be localized using @samp{_(@var{msg})}.
9888 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9889 when the PCH file was created and @var{sz} is the size of that data in bytes.
9890 It's safe to assume that the data was created by the same version of the
9891 compiler, so no format checking is needed.
9893 The default definition of @code{default_pch_valid_p} should be
9894 suitable for most targets.
9897 @hook TARGET_CHECK_PCH_TARGET_FLAGS
9898 If this hook is nonnull, the default implementation of
9899 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9900 of @code{target_flags}. @var{pch_flags} specifies the value that
9901 @code{target_flags} had when the PCH file was created. The return
9902 value is the same as for @code{TARGET_PCH_VALID_P}.
9906 @section C++ ABI parameters
9907 @cindex parameters, c++ abi
9909 @hook TARGET_CXX_GUARD_TYPE
9910 Define this hook to override the integer type used for guard variables.
9911 These are used to implement one-time construction of static objects. The
9912 default is long_long_integer_type_node.
9915 @hook TARGET_CXX_GUARD_MASK_BIT
9916 This hook determines how guard variables are used. It should return
9917 @code{false} (the default) if the first byte should be used. A return value of
9918 @code{true} indicates that only the least significant bit should be used.
9921 @hook TARGET_CXX_GET_COOKIE_SIZE
9922 This hook returns the size of the cookie to use when allocating an array
9923 whose elements have the indicated @var{type}. Assumes that it is already
9924 known that a cookie is needed. The default is
9925 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
9926 IA64/Generic C++ ABI@.
9929 @hook TARGET_CXX_COOKIE_HAS_SIZE
9930 This hook should return @code{true} if the element size should be stored in
9931 array cookies. The default is to return @code{false}.
9934 @hook TARGET_CXX_IMPORT_EXPORT_CLASS
9935 If defined by a backend this hook allows the decision made to export
9936 class @var{type} to be overruled. Upon entry @var{import_export}
9937 will contain 1 if the class is going to be exported, @minus{}1 if it is going
9938 to be imported and 0 otherwise. This function should return the
9939 modified value and perform any other actions necessary to support the
9940 backend's targeted operating system.
9943 @hook TARGET_CXX_CDTOR_RETURNS_THIS
9944 This hook should return @code{true} if constructors and destructors return
9945 the address of the object created/destroyed. The default is to return
9949 @hook TARGET_CXX_KEY_METHOD_MAY_BE_INLINE
9950 This hook returns true if the key method for a class (i.e., the method
9951 which, if defined in the current translation unit, causes the virtual
9952 table to be emitted) may be an inline function. Under the standard
9953 Itanium C++ ABI the key method may be an inline function so long as
9954 the function is not declared inline in the class definition. Under
9955 some variants of the ABI, an inline function can never be the key
9956 method. The default is to return @code{true}.
9959 @hook TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY
9961 @hook TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT
9962 This hook returns true (the default) if virtual tables and other
9963 similar implicit class data objects are always COMDAT if they have
9964 external linkage. If this hook returns false, then class data for
9965 classes whose virtual table will be emitted in only one translation
9966 unit will not be COMDAT.
9969 @hook TARGET_CXX_LIBRARY_RTTI_COMDAT
9970 This hook returns true (the default) if the RTTI information for
9971 the basic types which is defined in the C++ runtime should always
9972 be COMDAT, false if it should not be COMDAT.
9975 @hook TARGET_CXX_USE_AEABI_ATEXIT
9976 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
9977 should be used to register static destructors when @option{-fuse-cxa-atexit}
9978 is in effect. The default is to return false to use @code{__cxa_atexit}.
9981 @hook TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT
9982 This hook returns true if the target @code{atexit} function can be used
9983 in the same manner as @code{__cxa_atexit} to register C++ static
9984 destructors. This requires that @code{atexit}-registered functions in
9985 shared libraries are run in the correct order when the libraries are
9986 unloaded. The default is to return false.
9989 @hook TARGET_CXX_ADJUST_CLASS_AT_DEFINITION
9991 @node Named Address Spaces
9992 @section Adding support for named address spaces
9993 @cindex named address spaces
9995 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
9996 standards committee, @cite{Programming Languages - C - Extensions to
9997 support embedded processors}, specifies a syntax for embedded
9998 processors to specify alternate address spaces. You can configure a
9999 GCC port to support section 5.1 of the draft report to add support for
10000 address spaces other than the default address space. These address
10001 spaces are new keywords that are similar to the @code{volatile} and
10002 @code{const} type attributes.
10004 Pointers to named address spaces can have a different size than
10005 pointers to the generic address space.
10007 For example, the SPU port uses the @code{__ea} address space to refer
10008 to memory in the host processor, rather than memory local to the SPU
10009 processor. Access to memory in the @code{__ea} address space involves
10010 issuing DMA operations to move data between the host processor and the
10011 local processor memory address space. Pointers in the @code{__ea}
10012 address space are either 32 bits or 64 bits based on the
10013 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
10016 Internally, address spaces are represented as a small integer in the
10017 range 0 to 15 with address space 0 being reserved for the generic
10020 To register a named address space qualifier keyword with the C front end,
10021 the target may call the @code{c_register_addr_space} routine. For example,
10022 the SPU port uses the following to declare @code{__ea} as the keyword for
10023 named address space #1:
10025 #define ADDR_SPACE_EA 1
10026 c_register_addr_space ("__ea", ADDR_SPACE_EA);
10029 @hook TARGET_ADDR_SPACE_POINTER_MODE
10030 Define this to return the machine mode to use for pointers to
10031 @var{address_space} if the target supports named address spaces.
10032 The default version of this hook returns @code{ptr_mode} for the
10033 generic address space only.
10036 @hook TARGET_ADDR_SPACE_ADDRESS_MODE
10037 Define this to return the machine mode to use for addresses in
10038 @var{address_space} if the target supports named address spaces.
10039 The default version of this hook returns @code{Pmode} for the
10040 generic address space only.
10043 @hook TARGET_ADDR_SPACE_VALID_POINTER_MODE
10044 Define this to return nonzero if the port can handle pointers
10045 with machine mode @var{mode} to address space @var{as}. This target
10046 hook is the same as the @code{TARGET_VALID_POINTER_MODE} target hook,
10047 except that it includes explicit named address space support. The default
10048 version of this hook returns true for the modes returned by either the
10049 @code{TARGET_ADDR_SPACE_POINTER_MODE} or @code{TARGET_ADDR_SPACE_ADDRESS_MODE}
10050 target hooks for the given address space.
10053 @hook TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P
10054 Define this to return true if @var{exp} is a valid address for mode
10055 @var{mode} in the named address space @var{as}. The @var{strict}
10056 parameter says whether strict addressing is in effect after reload has
10057 finished. This target hook is the same as the
10058 @code{TARGET_LEGITIMATE_ADDRESS_P} target hook, except that it includes
10059 explicit named address space support.
10062 @hook TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS
10063 Define this to modify an invalid address @var{x} to be a valid address
10064 with mode @var{mode} in the named address space @var{as}. This target
10065 hook is the same as the @code{TARGET_LEGITIMIZE_ADDRESS} target hook,
10066 except that it includes explicit named address space support.
10069 @hook TARGET_ADDR_SPACE_SUBSET_P
10070 Define this to return whether the @var{subset} named address space is
10071 contained within the @var{superset} named address space. Pointers to
10072 a named address space that is a subset of another named address space
10073 will be converted automatically without a cast if used together in
10074 arithmetic operations. Pointers to a superset address space can be
10075 converted to pointers to a subset address space via explicit casts.
10078 @hook TARGET_ADDR_SPACE_CONVERT
10079 Define this to convert the pointer expression represented by the RTL
10080 @var{op} with type @var{from_type} that points to a named address
10081 space to a new pointer expression with type @var{to_type} that points
10082 to a different named address space. When this hook it called, it is
10083 guaranteed that one of the two address spaces is a subset of the other,
10084 as determined by the @code{TARGET_ADDR_SPACE_SUBSET_P} target hook.
10088 @section Miscellaneous Parameters
10089 @cindex parameters, miscellaneous
10091 @c prevent bad page break with this line
10092 Here are several miscellaneous parameters.
10094 @defmac HAS_LONG_COND_BRANCH
10095 Define this boolean macro to indicate whether or not your architecture
10096 has conditional branches that can span all of memory. It is used in
10097 conjunction with an optimization that partitions hot and cold basic
10098 blocks into separate sections of the executable. If this macro is
10099 set to false, gcc will convert any conditional branches that attempt
10100 to cross between sections into unconditional branches or indirect jumps.
10103 @defmac HAS_LONG_UNCOND_BRANCH
10104 Define this boolean macro to indicate whether or not your architecture
10105 has unconditional branches that can span all of memory. It is used in
10106 conjunction with an optimization that partitions hot and cold basic
10107 blocks into separate sections of the executable. If this macro is
10108 set to false, gcc will convert any unconditional branches that attempt
10109 to cross between sections into indirect jumps.
10112 @defmac CASE_VECTOR_MODE
10113 An alias for a machine mode name. This is the machine mode that
10114 elements of a jump-table should have.
10117 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
10118 Optional: return the preferred mode for an @code{addr_diff_vec}
10119 when the minimum and maximum offset are known. If you define this,
10120 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
10121 To make this work, you also have to define @code{INSN_ALIGN} and
10122 make the alignment for @code{addr_diff_vec} explicit.
10123 The @var{body} argument is provided so that the offset_unsigned and scale
10124 flags can be updated.
10127 @defmac CASE_VECTOR_PC_RELATIVE
10128 Define this macro to be a C expression to indicate when jump-tables
10129 should contain relative addresses. You need not define this macro if
10130 jump-tables never contain relative addresses, or jump-tables should
10131 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
10135 @hook TARGET_CASE_VALUES_THRESHOLD
10136 This function return the smallest number of different values for which it
10137 is best to use a jump-table instead of a tree of conditional branches.
10138 The default is four for machines with a @code{casesi} instruction and
10139 five otherwise. This is best for most machines.
10142 @defmac CASE_USE_BIT_TESTS
10143 Define this macro to be a C expression to indicate whether C switch
10144 statements may be implemented by a sequence of bit tests. This is
10145 advantageous on processors that can efficiently implement left shift
10146 of 1 by the number of bits held in a register, but inappropriate on
10147 targets that would require a loop. By default, this macro returns
10148 @code{true} if the target defines an @code{ashlsi3} pattern, and
10149 @code{false} otherwise.
10152 @defmac WORD_REGISTER_OPERATIONS
10153 Define this macro if operations between registers with integral mode
10154 smaller than a word are always performed on the entire register.
10155 Most RISC machines have this property and most CISC machines do not.
10158 @defmac LOAD_EXTEND_OP (@var{mem_mode})
10159 Define this macro to be a C expression indicating when insns that read
10160 memory in @var{mem_mode}, an integral mode narrower than a word, set the
10161 bits outside of @var{mem_mode} to be either the sign-extension or the
10162 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
10163 of @var{mem_mode} for which the
10164 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
10165 @code{UNKNOWN} for other modes.
10167 This macro is not called with @var{mem_mode} non-integral or with a width
10168 greater than or equal to @code{BITS_PER_WORD}, so you may return any
10169 value in this case. Do not define this macro if it would always return
10170 @code{UNKNOWN}. On machines where this macro is defined, you will normally
10171 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
10173 You may return a non-@code{UNKNOWN} value even if for some hard registers
10174 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
10175 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
10176 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
10177 integral mode larger than this but not larger than @code{word_mode}.
10179 You must return @code{UNKNOWN} if for some hard registers that allow this
10180 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
10181 @code{word_mode}, but that they can change to another integral mode that
10182 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
10185 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
10186 Define this macro if loading short immediate values into registers sign
10190 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
10191 Define this macro if the same instructions that convert a floating
10192 point number to a signed fixed point number also convert validly to an
10196 @hook TARGET_MIN_DIVISIONS_FOR_RECIP_MUL
10197 When @option{-ffast-math} is in effect, GCC tries to optimize
10198 divisions by the same divisor, by turning them into multiplications by
10199 the reciprocal. This target hook specifies the minimum number of divisions
10200 that should be there for GCC to perform the optimization for a variable
10201 of mode @var{mode}. The default implementation returns 3 if the machine
10202 has an instruction for the division, and 2 if it does not.
10206 The maximum number of bytes that a single instruction can move quickly
10207 between memory and registers or between two memory locations.
10210 @defmac MAX_MOVE_MAX
10211 The maximum number of bytes that a single instruction can move quickly
10212 between memory and registers or between two memory locations. If this
10213 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
10214 constant value that is the largest value that @code{MOVE_MAX} can have
10218 @defmac SHIFT_COUNT_TRUNCATED
10219 A C expression that is nonzero if on this machine the number of bits
10220 actually used for the count of a shift operation is equal to the number
10221 of bits needed to represent the size of the object being shifted. When
10222 this macro is nonzero, the compiler will assume that it is safe to omit
10223 a sign-extend, zero-extend, and certain bitwise `and' instructions that
10224 truncates the count of a shift operation. On machines that have
10225 instructions that act on bit-fields at variable positions, which may
10226 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
10227 also enables deletion of truncations of the values that serve as
10228 arguments to bit-field instructions.
10230 If both types of instructions truncate the count (for shifts) and
10231 position (for bit-field operations), or if no variable-position bit-field
10232 instructions exist, you should define this macro.
10234 However, on some machines, such as the 80386 and the 680x0, truncation
10235 only applies to shift operations and not the (real or pretended)
10236 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
10237 such machines. Instead, add patterns to the @file{md} file that include
10238 the implied truncation of the shift instructions.
10240 You need not define this macro if it would always have the value of zero.
10243 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
10244 @hook TARGET_SHIFT_TRUNCATION_MASK
10245 This function describes how the standard shift patterns for @var{mode}
10246 deal with shifts by negative amounts or by more than the width of the mode.
10247 @xref{shift patterns}.
10249 On many machines, the shift patterns will apply a mask @var{m} to the
10250 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
10251 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
10252 this is true for mode @var{mode}, the function should return @var{m},
10253 otherwise it should return 0. A return value of 0 indicates that no
10254 particular behavior is guaranteed.
10256 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
10257 @emph{not} apply to general shift rtxes; it applies only to instructions
10258 that are generated by the named shift patterns.
10260 The default implementation of this function returns
10261 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
10262 and 0 otherwise. This definition is always safe, but if
10263 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
10264 nevertheless truncate the shift count, you may get better code
10268 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
10269 A C expression which is nonzero if on this machine it is safe to
10270 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
10271 bits (where @var{outprec} is smaller than @var{inprec}) by merely
10272 operating on it as if it had only @var{outprec} bits.
10274 On many machines, this expression can be 1.
10276 @c rearranged this, removed the phrase "it is reported that". this was
10277 @c to fix an overfull hbox. --mew 10feb93
10278 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
10279 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
10280 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
10281 such cases may improve things.
10284 @hook TARGET_MODE_REP_EXTENDED
10285 The representation of an integral mode can be such that the values
10286 are always extended to a wider integral mode. Return
10287 @code{SIGN_EXTEND} if values of @var{mode} are represented in
10288 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10289 otherwise. (Currently, none of the targets use zero-extended
10290 representation this way so unlike @code{LOAD_EXTEND_OP},
10291 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10292 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10293 @var{mode} to @var{rep_mode} so that @var{rep_mode} is not the next
10294 widest integral mode and currently we take advantage of this fact.)
10296 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10297 value even if the extension is not performed on certain hard registers
10298 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10299 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10301 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10302 describe two related properties. If you define
10303 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10304 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10307 In order to enforce the representation of @code{mode},
10308 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10312 @defmac STORE_FLAG_VALUE
10313 A C expression describing the value returned by a comparison operator
10314 with an integral mode and stored by a store-flag instruction
10315 (@samp{cstore@var{mode}4}) when the condition is true. This description must
10316 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
10317 comparison operators whose results have a @code{MODE_INT} mode.
10319 A value of 1 or @minus{}1 means that the instruction implementing the
10320 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10321 and 0 when the comparison is false. Otherwise, the value indicates
10322 which bits of the result are guaranteed to be 1 when the comparison is
10323 true. This value is interpreted in the mode of the comparison
10324 operation, which is given by the mode of the first operand in the
10325 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
10326 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10329 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10330 generate code that depends only on the specified bits. It can also
10331 replace comparison operators with equivalent operations if they cause
10332 the required bits to be set, even if the remaining bits are undefined.
10333 For example, on a machine whose comparison operators return an
10334 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10335 @samp{0x80000000}, saying that just the sign bit is relevant, the
10339 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10343 can be converted to
10346 (ashift:SI @var{x} (const_int @var{n}))
10350 where @var{n} is the appropriate shift count to move the bit being
10351 tested into the sign bit.
10353 There is no way to describe a machine that always sets the low-order bit
10354 for a true value, but does not guarantee the value of any other bits,
10355 but we do not know of any machine that has such an instruction. If you
10356 are trying to port GCC to such a machine, include an instruction to
10357 perform a logical-and of the result with 1 in the pattern for the
10358 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10360 Often, a machine will have multiple instructions that obtain a value
10361 from a comparison (or the condition codes). Here are rules to guide the
10362 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10367 Use the shortest sequence that yields a valid definition for
10368 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10369 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10370 comparison operators to do so because there may be opportunities to
10371 combine the normalization with other operations.
10374 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10375 slightly preferred on machines with expensive jumps and 1 preferred on
10379 As a second choice, choose a value of @samp{0x80000001} if instructions
10380 exist that set both the sign and low-order bits but do not define the
10384 Otherwise, use a value of @samp{0x80000000}.
10387 Many machines can produce both the value chosen for
10388 @code{STORE_FLAG_VALUE} and its negation in the same number of
10389 instructions. On those machines, you should also define a pattern for
10390 those cases, e.g., one matching
10393 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10396 Some machines can also perform @code{and} or @code{plus} operations on
10397 condition code values with less instructions than the corresponding
10398 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
10399 machines, define the appropriate patterns. Use the names @code{incscc}
10400 and @code{decscc}, respectively, for the patterns which perform
10401 @code{plus} or @code{minus} operations on condition code values. See
10402 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
10403 find such instruction sequences on other machines.
10405 If this macro is not defined, the default value, 1, is used. You need
10406 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10407 instructions, or if the value generated by these instructions is 1.
10410 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10411 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10412 returned when comparison operators with floating-point results are true.
10413 Define this macro on machines that have comparison operations that return
10414 floating-point values. If there are no such operations, do not define
10418 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10419 A C expression that gives a rtx representing the nonzero true element
10420 for vector comparisons. The returned rtx should be valid for the inner
10421 mode of @var{mode} which is guaranteed to be a vector mode. Define
10422 this macro on machines that have vector comparison operations that
10423 return a vector result. If there are no such operations, do not define
10424 this macro. Typically, this macro is defined as @code{const1_rtx} or
10425 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10426 the compiler optimizing such vector comparison operations for the
10430 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10431 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10432 A C expression that indicates whether the architecture defines a value
10433 for @code{clz} or @code{ctz} with a zero operand.
10434 A result of @code{0} indicates the value is undefined.
10435 If the value is defined for only the RTL expression, the macro should
10436 evaluate to @code{1}; if the value applies also to the corresponding optab
10437 entry (which is normally the case if it expands directly into
10438 the corresponding RTL), then the macro should evaluate to @code{2}.
10439 In the cases where the value is defined, @var{value} should be set to
10442 If this macro is not defined, the value of @code{clz} or
10443 @code{ctz} at zero is assumed to be undefined.
10445 This macro must be defined if the target's expansion for @code{ffs}
10446 relies on a particular value to get correct results. Otherwise it
10447 is not necessary, though it may be used to optimize some corner cases, and
10448 to provide a default expansion for the @code{ffs} optab.
10450 Note that regardless of this macro the ``definedness'' of @code{clz}
10451 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10452 visible to the user. Thus one may be free to adjust the value at will
10453 to match the target expansion of these operations without fear of
10458 An alias for the machine mode for pointers. On most machines, define
10459 this to be the integer mode corresponding to the width of a hardware
10460 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10461 On some machines you must define this to be one of the partial integer
10462 modes, such as @code{PSImode}.
10464 The width of @code{Pmode} must be at least as large as the value of
10465 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10466 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10470 @defmac FUNCTION_MODE
10471 An alias for the machine mode used for memory references to functions
10472 being called, in @code{call} RTL expressions. On most CISC machines,
10473 where an instruction can begin at any byte address, this should be
10474 @code{QImode}. On most RISC machines, where all instructions have fixed
10475 size and alignment, this should be a mode with the same size and alignment
10476 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10479 @defmac STDC_0_IN_SYSTEM_HEADERS
10480 In normal operation, the preprocessor expands @code{__STDC__} to the
10481 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10482 hosts, like Solaris, the system compiler uses a different convention,
10483 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10484 strict conformance to the C Standard.
10486 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10487 convention when processing system header files, but when processing user
10488 files @code{__STDC__} will always expand to 1.
10491 @defmac NO_IMPLICIT_EXTERN_C
10492 Define this macro if the system header files support C++ as well as C@.
10493 This macro inhibits the usual method of using system header files in
10494 C++, which is to pretend that the file's contents are enclosed in
10495 @samp{extern "C" @{@dots{}@}}.
10500 @defmac REGISTER_TARGET_PRAGMAS ()
10501 Define this macro if you want to implement any target-specific pragmas.
10502 If defined, it is a C expression which makes a series of calls to
10503 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10504 for each pragma. The macro may also do any
10505 setup required for the pragmas.
10507 The primary reason to define this macro is to provide compatibility with
10508 other compilers for the same target. In general, we discourage
10509 definition of target-specific pragmas for GCC@.
10511 If the pragma can be implemented by attributes then you should consider
10512 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10514 Preprocessor macros that appear on pragma lines are not expanded. All
10515 @samp{#pragma} directives that do not match any registered pragma are
10516 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10519 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10520 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10522 Each call to @code{c_register_pragma} or
10523 @code{c_register_pragma_with_expansion} establishes one pragma. The
10524 @var{callback} routine will be called when the preprocessor encounters a
10528 #pragma [@var{space}] @var{name} @dots{}
10531 @var{space} is the case-sensitive namespace of the pragma, or
10532 @code{NULL} to put the pragma in the global namespace. The callback
10533 routine receives @var{pfile} as its first argument, which can be passed
10534 on to cpplib's functions if necessary. You can lex tokens after the
10535 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10536 callback will be silently ignored. The end of the line is indicated by
10537 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10538 arguments of pragmas registered with
10539 @code{c_register_pragma_with_expansion} but not on the arguments of
10540 pragmas registered with @code{c_register_pragma}.
10542 Note that the use of @code{pragma_lex} is specific to the C and C++
10543 compilers. It will not work in the Java or Fortran compilers, or any
10544 other language compilers for that matter. Thus if @code{pragma_lex} is going
10545 to be called from target-specific code, it must only be done so when
10546 building the C and C++ compilers. This can be done by defining the
10547 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10548 target entry in the @file{config.gcc} file. These variables should name
10549 the target-specific, language-specific object file which contains the
10550 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10551 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10552 how to build this object file.
10557 @defmac HANDLE_SYSV_PRAGMA
10558 Define this macro (to a value of 1) if you want the System V style
10559 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
10560 [=<value>]} to be supported by gcc.
10562 The pack pragma specifies the maximum alignment (in bytes) of fields
10563 within a structure, in much the same way as the @samp{__aligned__} and
10564 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
10565 the behavior to the default.
10567 A subtlety for Microsoft Visual C/C++ style bit-field packing
10568 (e.g.@: -mms-bitfields) for targets that support it:
10569 When a bit-field is inserted into a packed record, the whole size
10570 of the underlying type is used by one or more same-size adjacent
10571 bit-fields (that is, if its long:3, 32 bits is used in the record,
10572 and any additional adjacent long bit-fields are packed into the same
10573 chunk of 32 bits. However, if the size changes, a new field of that
10574 size is allocated).
10576 If both MS bit-fields and @samp{__attribute__((packed))} are used,
10577 the latter will take precedence. If @samp{__attribute__((packed))} is
10578 used on a single field when MS bit-fields are in use, it will take
10579 precedence for that field, but the alignment of the rest of the structure
10580 may affect its placement.
10582 The weak pragma only works if @code{SUPPORTS_WEAK} and
10583 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
10584 of specifically named weak labels, optionally with a value.
10589 @defmac HANDLE_PRAGMA_PACK_PUSH_POP
10590 Define this macro (to a value of 1) if you want to support the Win32
10591 style pragmas @samp{#pragma pack(push[,@var{n}])} and @samp{#pragma
10592 pack(pop)}. The @samp{pack(push,[@var{n}])} pragma specifies the maximum
10593 alignment (in bytes) of fields within a structure, in much the same way as
10594 the @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
10595 pack value of zero resets the behavior to the default. Successive
10596 invocations of this pragma cause the previous values to be stacked, so
10597 that invocations of @samp{#pragma pack(pop)} will return to the previous
10601 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10602 Define this macro, as well as
10603 @code{HANDLE_SYSV_PRAGMA}, if macros should be expanded in the
10604 arguments of @samp{#pragma pack}.
10607 @hook TARGET_HANDLE_PRAGMA_EXTERN_PREFIX
10609 @defmac TARGET_DEFAULT_PACK_STRUCT
10610 If your target requires a structure packing default other than 0 (meaning
10611 the machine default), define this macro to the necessary value (in bytes).
10612 This must be a value that would also be valid to use with
10613 @samp{#pragma pack()} (that is, a small power of two).
10616 @defmac DOLLARS_IN_IDENTIFIERS
10617 Define this macro to control use of the character @samp{$} in
10618 identifier names for the C family of languages. 0 means @samp{$} is
10619 not allowed by default; 1 means it is allowed. 1 is the default;
10620 there is no need to define this macro in that case.
10623 @defmac NO_DOLLAR_IN_LABEL
10624 Define this macro if the assembler does not accept the character
10625 @samp{$} in label names. By default constructors and destructors in
10626 G++ have @samp{$} in the identifiers. If this macro is defined,
10627 @samp{.} is used instead.
10630 @defmac NO_DOT_IN_LABEL
10631 Define this macro if the assembler does not accept the character
10632 @samp{.} in label names. By default constructors and destructors in G++
10633 have names that use @samp{.}. If this macro is defined, these names
10634 are rewritten to avoid @samp{.}.
10637 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10638 Define this macro as a C expression that is nonzero if it is safe for the
10639 delay slot scheduler to place instructions in the delay slot of @var{insn},
10640 even if they appear to use a resource set or clobbered in @var{insn}.
10641 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10642 every @code{call_insn} has this behavior. On machines where some @code{insn}
10643 or @code{jump_insn} is really a function call and hence has this behavior,
10644 you should define this macro.
10646 You need not define this macro if it would always return zero.
10649 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10650 Define this macro as a C expression that is nonzero if it is safe for the
10651 delay slot scheduler to place instructions in the delay slot of @var{insn},
10652 even if they appear to set or clobber a resource referenced in @var{insn}.
10653 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10654 some @code{insn} or @code{jump_insn} is really a function call and its operands
10655 are registers whose use is actually in the subroutine it calls, you should
10656 define this macro. Doing so allows the delay slot scheduler to move
10657 instructions which copy arguments into the argument registers into the delay
10658 slot of @var{insn}.
10660 You need not define this macro if it would always return zero.
10663 @defmac MULTIPLE_SYMBOL_SPACES
10664 Define this macro as a C expression that is nonzero if, in some cases,
10665 global symbols from one translation unit may not be bound to undefined
10666 symbols in another translation unit without user intervention. For
10667 instance, under Microsoft Windows symbols must be explicitly imported
10668 from shared libraries (DLLs).
10670 You need not define this macro if it would always evaluate to zero.
10673 @hook TARGET_MD_ASM_CLOBBERS
10674 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10675 any hard regs the port wishes to automatically clobber for an asm.
10676 It should return the result of the last @code{tree_cons} used to add a
10677 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10678 corresponding parameters to the asm and may be inspected to avoid
10679 clobbering a register that is an input or output of the asm. You can use
10680 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10681 for overlap with regards to asm-declared registers.
10684 @defmac MATH_LIBRARY
10685 Define this macro as a C string constant for the linker argument to link
10686 in the system math library, or @samp{""} if the target does not have a
10687 separate math library.
10689 You need only define this macro if the default of @samp{"-lm"} is wrong.
10692 @defmac LIBRARY_PATH_ENV
10693 Define this macro as a C string constant for the environment variable that
10694 specifies where the linker should look for libraries.
10696 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10700 @defmac TARGET_POSIX_IO
10701 Define this macro if the target supports the following POSIX@ file
10702 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10703 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10704 to use file locking when exiting a program, which avoids race conditions
10705 if the program has forked. It will also create directories at run-time
10706 for cross-profiling.
10709 @defmac MAX_CONDITIONAL_EXECUTE
10711 A C expression for the maximum number of instructions to execute via
10712 conditional execution instructions instead of a branch. A value of
10713 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10714 1 if it does use cc0.
10717 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10718 Used if the target needs to perform machine-dependent modifications on the
10719 conditionals used for turning basic blocks into conditionally executed code.
10720 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10721 contains information about the currently processed blocks. @var{true_expr}
10722 and @var{false_expr} are the tests that are used for converting the
10723 then-block and the else-block, respectively. Set either @var{true_expr} or
10724 @var{false_expr} to a null pointer if the tests cannot be converted.
10727 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10728 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10729 if-statements into conditions combined by @code{and} and @code{or} operations.
10730 @var{bb} contains the basic block that contains the test that is currently
10731 being processed and about to be turned into a condition.
10734 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10735 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10736 be converted to conditional execution format. @var{ce_info} points to
10737 a data structure, @code{struct ce_if_block}, which contains information
10738 about the currently processed blocks.
10741 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10742 A C expression to perform any final machine dependent modifications in
10743 converting code to conditional execution. The involved basic blocks
10744 can be found in the @code{struct ce_if_block} structure that is pointed
10745 to by @var{ce_info}.
10748 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10749 A C expression to cancel any machine dependent modifications in
10750 converting code to conditional execution. The involved basic blocks
10751 can be found in the @code{struct ce_if_block} structure that is pointed
10752 to by @var{ce_info}.
10755 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
10756 A C expression to initialize any extra fields in a @code{struct ce_if_block}
10757 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
10760 @defmac IFCVT_EXTRA_FIELDS
10761 If defined, it should expand to a set of field declarations that will be
10762 added to the @code{struct ce_if_block} structure. These should be initialized
10763 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
10766 @hook TARGET_MACHINE_DEPENDENT_REORG
10767 If non-null, this hook performs a target-specific pass over the
10768 instruction stream. The compiler will run it at all optimization levels,
10769 just before the point at which it normally does delayed-branch scheduling.
10771 The exact purpose of the hook varies from target to target. Some use
10772 it to do transformations that are necessary for correctness, such as
10773 laying out in-function constant pools or avoiding hardware hazards.
10774 Others use it as an opportunity to do some machine-dependent optimizations.
10776 You need not implement the hook if it has nothing to do. The default
10777 definition is null.
10780 @hook TARGET_INIT_BUILTINS
10781 Define this hook if you have any machine-specific built-in functions
10782 that need to be defined. It should be a function that performs the
10785 Machine specific built-in functions can be useful to expand special machine
10786 instructions that would otherwise not normally be generated because
10787 they have no equivalent in the source language (for example, SIMD vector
10788 instructions or prefetch instructions).
10790 To create a built-in function, call the function
10791 @code{lang_hooks.builtin_function}
10792 which is defined by the language front end. You can use any type nodes set
10793 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
10794 only language front ends that use those two functions will call
10795 @samp{TARGET_INIT_BUILTINS}.
10798 @hook TARGET_BUILTIN_DECL
10799 Define this hook if you have any machine-specific built-in functions
10800 that need to be defined. It should be a function that returns the
10801 builtin function declaration for the builtin function code @var{code}.
10802 If there is no such builtin and it cannot be initialized at this time
10803 if @var{initialize_p} is true the function should return @code{NULL_TREE}.
10804 If @var{code} is out of range the function should return
10805 @code{error_mark_node}.
10808 @hook TARGET_EXPAND_BUILTIN
10810 Expand a call to a machine specific built-in function that was set up by
10811 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10812 function call; the result should go to @var{target} if that is
10813 convenient, and have mode @var{mode} if that is convenient.
10814 @var{subtarget} may be used as the target for computing one of
10815 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10816 ignored. This function should return the result of the call to the
10820 @hook TARGET_RESOLVE_OVERLOADED_BUILTIN
10821 Select a replacement for a machine specific built-in function that
10822 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10823 @emph{before} regular type checking, and so allows the target to
10824 implement a crude form of function overloading. @var{fndecl} is the
10825 declaration of the built-in function. @var{arglist} is the list of
10826 arguments passed to the built-in function. The result is a
10827 complete expression that implements the operation, usually
10828 another @code{CALL_EXPR}.
10829 @var{arglist} really has type @samp{VEC(tree,gc)*}
10832 @hook TARGET_FOLD_BUILTIN
10833 Fold a call to a machine specific built-in function that was set up by
10834 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10835 built-in function. @var{n_args} is the number of arguments passed to
10836 the function; the arguments themselves are pointed to by @var{argp}.
10837 The result is another tree containing a simplified expression for the
10838 call's result. If @var{ignore} is true the value will be ignored.
10841 @hook TARGET_INVALID_WITHIN_DOLOOP
10843 Take an instruction in @var{insn} and return NULL if it is valid within a
10844 low-overhead loop, otherwise return a string explaining why doloop
10845 could not be applied.
10847 Many targets use special registers for low-overhead looping. For any
10848 instruction that clobbers these this function should return a string indicating
10849 the reason why the doloop could not be applied.
10850 By default, the RTL loop optimizer does not use a present doloop pattern for
10851 loops containing function calls or branch on table instructions.
10854 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10856 Take a branch insn in @var{branch1} and another in @var{branch2}.
10857 Return true if redirecting @var{branch1} to the destination of
10858 @var{branch2} is possible.
10860 On some targets, branches may have a limited range. Optimizing the
10861 filling of delay slots can result in branches being redirected, and this
10862 may in turn cause a branch offset to overflow.
10865 @hook TARGET_COMMUTATIVE_P
10866 This target hook returns @code{true} if @var{x} is considered to be commutative.
10867 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10868 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10869 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10872 @hook TARGET_ALLOCATE_INITIAL_VALUE
10874 When the initial value of a hard register has been copied in a pseudo
10875 register, it is often not necessary to actually allocate another register
10876 to this pseudo register, because the original hard register or a stack slot
10877 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10878 is called at the start of register allocation once for each hard register
10879 that had its initial value copied by using
10880 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10881 Possible values are @code{NULL_RTX}, if you don't want
10882 to do any special allocation, a @code{REG} rtx---that would typically be
10883 the hard register itself, if it is known not to be clobbered---or a
10885 If you are returning a @code{MEM}, this is only a hint for the allocator;
10886 it might decide to use another register anyways.
10887 You may use @code{current_function_leaf_function} in the hook, functions
10888 that use @code{REG_N_SETS}, to determine if the hard
10889 register in question will not be clobbered.
10890 The default value of this hook is @code{NULL}, which disables any special
10894 @hook TARGET_UNSPEC_MAY_TRAP_P
10895 This target hook returns nonzero if @var{x}, an @code{unspec} or
10896 @code{unspec_volatile} operation, might cause a trap. Targets can use
10897 this hook to enhance precision of analysis for @code{unspec} and
10898 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
10899 to analyze inner elements of @var{x} in which case @var{flags} should be
10903 @hook TARGET_SET_CURRENT_FUNCTION
10904 The compiler invokes this hook whenever it changes its current function
10905 context (@code{cfun}). You can define this function if
10906 the back end needs to perform any initialization or reset actions on a
10907 per-function basis. For example, it may be used to implement function
10908 attributes that affect register usage or code generation patterns.
10909 The argument @var{decl} is the declaration for the new function context,
10910 and may be null to indicate that the compiler has left a function context
10911 and is returning to processing at the top level.
10912 The default hook function does nothing.
10914 GCC sets @code{cfun} to a dummy function context during initialization of
10915 some parts of the back end. The hook function is not invoked in this
10916 situation; you need not worry about the hook being invoked recursively,
10917 or when the back end is in a partially-initialized state.
10918 @code{cfun} might be @code{NULL} to indicate processing at top level,
10919 outside of any function scope.
10922 @defmac TARGET_OBJECT_SUFFIX
10923 Define this macro to be a C string representing the suffix for object
10924 files on your target machine. If you do not define this macro, GCC will
10925 use @samp{.o} as the suffix for object files.
10928 @defmac TARGET_EXECUTABLE_SUFFIX
10929 Define this macro to be a C string representing the suffix to be
10930 automatically added to executable files on your target machine. If you
10931 do not define this macro, GCC will use the null string as the suffix for
10935 @defmac COLLECT_EXPORT_LIST
10936 If defined, @code{collect2} will scan the individual object files
10937 specified on its command line and create an export list for the linker.
10938 Define this macro for systems like AIX, where the linker discards
10939 object files that are not referenced from @code{main} and uses export
10943 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
10944 Define this macro to a C expression representing a variant of the
10945 method call @var{mdecl}, if Java Native Interface (JNI) methods
10946 must be invoked differently from other methods on your target.
10947 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
10948 the @code{stdcall} calling convention and this macro is then
10949 defined as this expression:
10952 build_type_attribute_variant (@var{mdecl},
10954 (get_identifier ("stdcall"),
10959 @hook TARGET_CANNOT_MODIFY_JUMPS_P
10960 This target hook returns @code{true} past the point in which new jump
10961 instructions could be created. On machines that require a register for
10962 every jump such as the SHmedia ISA of SH5, this point would typically be
10963 reload, so this target hook should be defined to a function such as:
10967 cannot_modify_jumps_past_reload_p ()
10969 return (reload_completed || reload_in_progress);
10974 @hook TARGET_BRANCH_TARGET_REGISTER_CLASS
10975 This target hook returns a register class for which branch target register
10976 optimizations should be applied. All registers in this class should be
10977 usable interchangeably. After reload, registers in this class will be
10978 re-allocated and loads will be hoisted out of loops and be subjected
10979 to inter-block scheduling.
10982 @hook TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED
10983 Branch target register optimization will by default exclude callee-saved
10985 that are not already live during the current function; if this target hook
10986 returns true, they will be included. The target code must than make sure
10987 that all target registers in the class returned by
10988 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
10989 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
10990 epilogues have already been generated. Note, even if you only return
10991 true when @var{after_prologue_epilogue_gen} is false, you still are likely
10992 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
10993 to reserve space for caller-saved target registers.
10996 @hook TARGET_HAVE_CONDITIONAL_EXECUTION
10997 This target hook returns true if the target supports conditional execution.
10998 This target hook is required only when the target has several different
10999 modes and they have different conditional execution capability, such as ARM.
11002 @hook TARGET_LOOP_UNROLL_ADJUST
11003 This target hook returns a new value for the number of times @var{loop}
11004 should be unrolled. The parameter @var{nunroll} is the number of times
11005 the loop is to be unrolled. The parameter @var{loop} is a pointer to
11006 the loop, which is going to be checked for unrolling. This target hook
11007 is required only when the target has special constraints like maximum
11008 number of memory accesses.
11011 @defmac POWI_MAX_MULTS
11012 If defined, this macro is interpreted as a signed integer C expression
11013 that specifies the maximum number of floating point multiplications
11014 that should be emitted when expanding exponentiation by an integer
11015 constant inline. When this value is defined, exponentiation requiring
11016 more than this number of multiplications is implemented by calling the
11017 system library's @code{pow}, @code{powf} or @code{powl} routines.
11018 The default value places no upper bound on the multiplication count.
11021 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11022 This target hook should register any extra include files for the
11023 target. The parameter @var{stdinc} indicates if normal include files
11024 are present. The parameter @var{sysroot} is the system root directory.
11025 The parameter @var{iprefix} is the prefix for the gcc directory.
11028 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11029 This target hook should register any extra include files for the
11030 target before any standard headers. The parameter @var{stdinc}
11031 indicates if normal include files are present. The parameter
11032 @var{sysroot} is the system root directory. The parameter
11033 @var{iprefix} is the prefix for the gcc directory.
11036 @deftypefn Macro void TARGET_OPTF (char *@var{path})
11037 This target hook should register special include paths for the target.
11038 The parameter @var{path} is the include to register. On Darwin
11039 systems, this is used for Framework includes, which have semantics
11040 that are different from @option{-I}.
11043 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
11044 This target macro returns @code{true} if it is safe to use a local alias
11045 for a virtual function @var{fndecl} when constructing thunks,
11046 @code{false} otherwise. By default, the macro returns @code{true} for all
11047 functions, if a target supports aliases (i.e.@: defines
11048 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
11051 @defmac TARGET_FORMAT_TYPES
11052 If defined, this macro is the name of a global variable containing
11053 target-specific format checking information for the @option{-Wformat}
11054 option. The default is to have no target-specific format checks.
11057 @defmac TARGET_N_FORMAT_TYPES
11058 If defined, this macro is the number of entries in
11059 @code{TARGET_FORMAT_TYPES}.
11062 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
11063 If defined, this macro is the name of a global variable containing
11064 target-specific format overrides for the @option{-Wformat} option. The
11065 default is to have no target-specific format overrides. If defined,
11066 @code{TARGET_FORMAT_TYPES} must be defined, too.
11069 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
11070 If defined, this macro specifies the number of entries in
11071 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
11074 @defmac TARGET_OVERRIDES_FORMAT_INIT
11075 If defined, this macro specifies the optional initialization
11076 routine for target specific customizations of the system printf
11077 and scanf formatter settings.
11080 @hook TARGET_RELAXED_ORDERING
11081 If set to @code{true}, means that the target's memory model does not
11082 guarantee that loads which do not depend on one another will access
11083 main memory in the order of the instruction stream; if ordering is
11084 important, an explicit memory barrier must be used. This is true of
11085 many recent processors which implement a policy of ``relaxed,''
11086 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
11087 and ia64. The default is @code{false}.
11090 @hook TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN
11091 If defined, this macro returns the diagnostic message when it is
11092 illegal to pass argument @var{val} to function @var{funcdecl}
11093 with prototype @var{typelist}.
11096 @hook TARGET_INVALID_CONVERSION
11097 If defined, this macro returns the diagnostic message when it is
11098 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
11099 if validity should be determined by the front end.
11102 @hook TARGET_INVALID_UNARY_OP
11103 If defined, this macro returns the diagnostic message when it is
11104 invalid to apply operation @var{op} (where unary plus is denoted by
11105 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
11106 if validity should be determined by the front end.
11109 @hook TARGET_INVALID_BINARY_OP
11110 If defined, this macro returns the diagnostic message when it is
11111 invalid to apply operation @var{op} to operands of types @var{type1}
11112 and @var{type2}, or @code{NULL} if validity should be determined by
11116 @hook TARGET_INVALID_PARAMETER_TYPE
11117 If defined, this macro returns the diagnostic message when it is
11118 invalid for functions to include parameters of type @var{type},
11119 or @code{NULL} if validity should be determined by
11120 the front end. This is currently used only by the C and C++ front ends.
11123 @hook TARGET_INVALID_RETURN_TYPE
11124 If defined, this macro returns the diagnostic message when it is
11125 invalid for functions to have return type @var{type},
11126 or @code{NULL} if validity should be determined by
11127 the front end. This is currently used only by the C and C++ front ends.
11130 @hook TARGET_PROMOTED_TYPE
11131 If defined, this target hook returns the type to which values of
11132 @var{type} should be promoted when they appear in expressions,
11133 analogous to the integer promotions, or @code{NULL_TREE} to use the
11134 front end's normal promotion rules. This hook is useful when there are
11135 target-specific types with special promotion rules.
11136 This is currently used only by the C and C++ front ends.
11139 @hook TARGET_CONVERT_TO_TYPE
11140 If defined, this hook returns the result of converting @var{expr} to
11141 @var{type}. It should return the converted expression,
11142 or @code{NULL_TREE} to apply the front end's normal conversion rules.
11143 This hook is useful when there are target-specific types with special
11145 This is currently used only by the C and C++ front ends.
11148 @defmac TARGET_USE_JCR_SECTION
11149 This macro determines whether to use the JCR section to register Java
11150 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
11151 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
11155 This macro determines the size of the objective C jump buffer for the
11156 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
11159 @defmac LIBGCC2_UNWIND_ATTRIBUTE
11160 Define this macro if any target-specific attributes need to be attached
11161 to the functions in @file{libgcc} that provide low-level support for
11162 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
11163 and the associated definitions of those functions.
11166 @hook TARGET_UPDATE_STACK_BOUNDARY
11167 Define this macro to update the current function stack boundary if
11171 @hook TARGET_GET_DRAP_RTX
11172 This hook should return an rtx for Dynamic Realign Argument Pointer (DRAP) if a
11173 different argument pointer register is needed to access the function's
11174 argument list due to stack realignment. Return @code{NULL} if no DRAP
11178 @hook TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS
11179 When optimization is disabled, this hook indicates whether or not
11180 arguments should be allocated to stack slots. Normally, GCC allocates
11181 stacks slots for arguments when not optimizing in order to make
11182 debugging easier. However, when a function is declared with
11183 @code{__attribute__((naked))}, there is no stack frame, and the compiler
11184 cannot safely move arguments from the registers in which they are passed
11185 to the stack. Therefore, this hook should return true in general, but
11186 false for naked functions. The default implementation always returns true.
11189 @hook TARGET_CONST_ANCHOR
11190 On some architectures it can take multiple instructions to synthesize
11191 a constant. If there is another constant already in a register that
11192 is close enough in value then it is preferable that the new constant
11193 is computed from this register using immediate addition or
11194 subtraction. We accomplish this through CSE. Besides the value of
11195 the constant we also add a lower and an upper constant anchor to the
11196 available expressions. These are then queried when encountering new
11197 constants. The anchors are computed by rounding the constant up and
11198 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
11199 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
11200 accepted by immediate-add plus one. We currently assume that the
11201 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
11202 MIPS, where add-immediate takes a 16-bit signed value,
11203 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
11204 is zero, which disables this optimization. @end deftypevr